Power supply for voltage references

[branadic]

Hi all,

triggered by the W/F7000 I thought it would be interesting to talk about power supplies for voltage references. Battery power is indeed a proper solution and Andreas showed a design that provides the voltage out of a set of 12x NiMH for the LTZ reference via an LT1763, while applied with BMON (battery monitor) a charge circuitry is present.
But what about mains power, with extrem low coupling capacity?

John Pickering obviously developed a DC-DC converter solution that seems to work pretty well. So I was looking for info about it on the web and came across the following papers, describing the solution. Carefully reading all three of them about the xformer together with the pictures of W7000 available on xdevs provides an overall picture how it's done:

Design of a 10 mA DC current reference standard
Design and evaluation of a 10-mA DC current reference standard
A new compact isolated power supply for electrical Metrology at low signal levels



So I wonder if anyone else created a similar solution, providing high isolation between primary and secondary, so that it is possible to power the reference from mains. What xformer was used?

Edit: Guidance on eliminating interference from sensitive electrical circuits

Edit: Design and Evaluation of a 10-mA DC Current Reference Standard states a frequency of 8 kHz for the DC-DC converter, which agrees with the frequency I can measure on my F7000 based on LT1533

-branadic-

[Kleinstein]

For extremely low capacitance / high voltage, there are acoustic transformers (piezo - isolator - piezo). However they are relatively exotic and usually for low power (like 1 W).

The usual small higher frequency transformers work better with a low voltage. So this would likely be in 2 stages: 1st a conventional mains transformer to some 5 V DC and than a special, low capacitance ferrite or similar transformer. Dave sowed a nice example from the Clever-scope.

If efficiency is not that important, on can have the primary and secondary windings on opposite sides of a relatively large ring core.

[exe]

Am I correct that you want a tranny with low capacitance between windings to avoid getting noise from mains?

R-core transformers?  I resist buying one, although I may pull the trigger. Alternatively, a toroid with a shield between primary and secondary might also be an option. May be even a conventional EI transformer with a gap between windings would suit you?

Anyway, even if we eliminate capacitance between windings, magentic coupling is still quite strong up to some tens of kHz or even beyond. I know this because people use small EI-trannies as injection transformers.

[doktor pyta]

Yup.
Resonant current fed push-pull operating at about 20kHz.
As far as I remember parameters were:
Supply:12 V, output 16V @0.3A. Efficiency 50-60%.
Capacitance C(in-out)=4pF.
Pros: very easy to build with commercial components, I've never observed problems with start-up.

[BravoV]

For extremely low capacitance / high voltage, there are acoustic transformers (piezo - isolator - piezo). However they are relatively exotic and usually for low power (like 1 W).

The usual small higher frequency transformers work better with a low voltage. So this would likely be in 2 stages: 1st a conventional mains transformer to some 5 V DC and than a special, low capacitance ferrite or similar transformer. Dave sowed a nice example from the Clever-scope.

If efficiency is not that important, on can have the primary and secondary windings on opposite sides of a relatively large ring core.

This one ?

[tszaboo]

I have two ideas for you:
- multi stage transformer (230 to 48V, 48 to 12)
- Jim Williams magic:
https://www.analog.com/media/en/technical-documentation/application-notes/an29f.pdf

There is a circuit, where he drives a transformer with a 1-2 KHz sine wave that comes from a current boosted opamp. Crap efficiency, but small and not noisy, so who cares.

[notfaded1]

Does anyone know why they refer in the IEEE paper to "The value of 10 mA, based on engineering considerations, seemed suitable for all of the following applications, allowing a single design of dc current reference, referred to as the PBC, to cater for each."  Anyone have any idea why they called them PBC?  Just curious.

Another couple questions... maybe dumb but it says "The five prototype units were each connected to a 100Ω resistor and measured daily with an MIL 8000 against Fluke 732B standards traceable to METAS/Bern. The long term drift can be seen in Fig. 2. When the measurement resistors were replaced by 1 kΩ , the inter-comparison could be made directly with the Fluke 732B 10-V standards through the DataProof scanner and uncertainty reduced. It is estimated that a group of four units can easily provide the 10 mA with an uncertainty of five parts in 10^7 constantly maintained.

What's a MIL 8000 and a DataProof scanner?  I'm guessing the DataProof scanner comes from here: http://www.dataproof.com/

Bill

[Vgkid]

Measurements international binary voltage divider
https://mintl.com/products/model-8000b-10-volt/

[Conrad Hoffman]

The construction details of the GR shielded bridge transformers are interesting- https://www.ietlabs.com/pdf/GR_Experimenters/1935/GenRad_Experimenter_Oct_1935.pdf
You can also shield a toroid with copper tape.
Me, I'd probably use a couple solar cells, lit up with some big incandescent lamp! Can't get much better isolation than that.

[dietert1]

After experiments with high frequency converters (art of electronics) i am now using something very simple and old fashioned. It's based on a small two chamber mains transformer with low coupling (20 pF) plus two extra 2K2 resistors on the input side (RF suppression). Like all two chamber transformers the secondary needs a proper snubber. Then there is a LM723 regulator with a 30 mA current limit. Outputs are adjusted to 15.000 V with fixed resistors. Temperature drift remains below +/- 3 mV in our lab. Image shows a 750 Ohm resistor i used for test (20 mA at 15 V). This circuit will inject less than 1 uArms current at 50 Hz into the ground line, so with reasonable wiring errors will be a fraction of a microvolt.

I found that our LTFLU references shifted by about 0.15 ppm when i replaced the previous LM7815 based power supplies (e.g. from 14.79 to 15.000 V). That means even a well designed reference will change when you change the supply voltage by such an amount.

Regards, Dieter

PS: I made six of those and have parts for more. Open two chamber transformers where you can ground the core are superior, but the minimum size i know of is 12 VA (much bigger) and the difference in performance is small.

[babysitter]

I digged trough a old Siemens patient monitor with exchangeable modules, they used a IR communication and two pot-core ferrites for power transmission, mounted like "E  3" . Think inductive charger. They did it for low leakage and cleanable surface, but should give low capacitance too. Maybe a royer circuit would be a good start for the primary part. Load the secondary to prevent high output voltage.

[SilverSolder]

The construction details of the GR shielded bridge transformers are interesting- https://www.ietlabs.com/pdf/GR_Experimenters/1935/GenRad_Experimenter_Oct_1935.pdf
You can also shield a toroid with copper tape.
Me, I'd probably use a couple solar cells, lit up with some big incandescent lamp! Can't get much better isolation than that.

I have a couple of double shielded tranformers made by Singer/Gertsch and they are also 1:4.

What is the reason for these double-shielded transformers often being made with a 1:4 turns ratio?

[branadic]

I found that our LTFLU references shifted by about 0.15 ppm when i replaced the previous LM7815 based power supplies (e.g. from 14.79 to 15.000 V). That means even a well designed reference will change when you change the supply voltage by such an amount.

I must add that W/F7000, F7001, F7004 and F7010 are powered by a Hitron Model HER-57-312 wall adapter with 12.88V and 1.2A. This powers the DC-DC converter, so it's already kind of a two stage concept.
Attached is a measurement of the averaged output on F7000T, running the modules from their internal batteries in the first half and from the wall adapter in the second half. As far as I can see there is absolutely no difference. One can repeat that of coarse on the individual units too, just in case there was some cancelling effect going on.

-branadic-

[niner_007]

I was thinking of starting a thread on it the other day

My thoughts on this are to use rechargeable batteries, stored in the same heavily shielded metal enclosure, alongside the reference, like in a Fluke 732. For the batteries, something practical with lots of energy and long life, like LiFePo or LiIon. I do not know what kind of noise these have, if lead acid has lower noise, that can be an option too. Keep two independent battery banks, one that is being charged, and one that is used to power the reference. As one bank is completely discharged, you switch over to the other bank. That way you remove any noise from ever entering the system from the outside, with runs with the assumption that the batteries are relatively low noise.

For regulation or step down to a practical range, I was thinking of a simple npn - zener regulator, using a low noise zener diode (1N5518, 1N5519, 1N5520, 1N5521, 1N5522, 1N5523). Another option that could be great to try is using the LTC6655 low noise reference. It could also be used to reduce the noise of the LT3042, which on its own has something like 10uV of noise 0.1 to 10Hz, and 2uV of noise 10Hz to 100kHz.

[TiN]

For the batteries, something practical with lots of energy and long life, like LiFePo or LiIon.

Ouch, that means good bye shipping for calibration without paperwork/etc.

[dietert1]

I liked the Cleverscope transformer, thanks. Using a twisted pair bifilar transformer with symmetrical drive in resonant mode (like in the art of electronics proposal) may also remain below 1 uArms of current injection. It's just a matter of designing a 3D printed part to hold off the secondary. That's something to try. I still have the circuit somewhere on the bench.

The AD paper AN29 mentions 30 uV somewhere, that's a lot. And high frequency is more difficult to absorb into wiring due to cable inductance.

How to measure the injected signal? Use two scope channels, one to trigger on the converter frequency (proximity, ground lead remains open), the other one to measure injected signal over a 1 KOhm resistor. One side of the resistor gets connected to converter output ground and probe tip, the other side gets connected to the ground lead of that probe. Then use averaging to raise injected signal above noise. With a modern DSO this will get you down to some 100 or 500 uV sensitivity, which then corresponds to about 0.1 to 0.5 uA injected current. Need to avoid clipping though: The scope power supply also injects its own asynchronous noise. Clipping will screw up averaging, so you may have to adjust scope sensitivity before.

Regards, Dieter

[notfaded1]

@niner_007  Another nice thing about using two sets of rechargeable batteries is you can change out the one side with new batteries while the other side is running.  You'd never have to come off of battery power that way ever and always have good fresh batteries.

Bill

[branadic]

Repeated the measurement battery vs. wall adapter this time on cell1 only, I leave it up to you what you can spot.

-branadic-

[notfaded1]

It's definitely doesn't look as stable as before.  Downward slope.

[Conrad Hoffman]

The construction details of the GR shielded bridge transformers are interesting- https://www.ietlabs.com/pdf/GR_Experimenters/1935/GenRad_Experimenter_Oct_1935.pdf
You can also shield a toroid with copper tape.
Me, I'd probably use a couple solar cells, lit up with some big incandescent lamp! Can't get much better isolation than that.

I have a couple of double shielded tranformers made by Singer/Gertsch and they are also 1:4.

What is the reason for these double-shielded transformers often being made with a 1:4 turns ratio?

Not a clue, but no doubt there was some sensible reason, possibly a desirable impedance conversion.

[niner_007]

For the batteries, something practical with lots of energy and long life, like LiFePo or LiIon.

Ouch, that means good bye shipping for calibration without paperwork/etc.

Can't have everything in life

You can build it modular, if you don't want batteries, it can be mains power. You can enable any chemistries also, and any capacities. Volumetric energy storage is what I mean, the batteries can be as low as 0.5Ah, it all depends on how long you want the system to run without external power.

[Jay_Diddy_B]

Hi,

The LT1533 is well suited to these applications. It has controlled slew rate for both current and voltage. This allows you to trade efficiency for low noise.

Link: https://www.analog.com/en/products/lt1533.html

If you don't need isolation, you can look at the very similar LT1534.

Regards,
Jay_Diddy_B

[dietert1]

The LT1533 specification of the injected signal seems to be 100 uVpp, including high frequency transients as visible on the scope trace, so once more a lot. That is because they are using switches, not linear drive as in the art of electronics proposal. Of course it also depends on the transformer geometry.

I remember from my experiments with the art of electronics circuit, that depending on the transformer the injected signal at the secondary ground could be between 300 uArms and 1 uArms. It got lower with lower frequency e.g. 20 KHz instead of 50 KHz and when using a slotted copper foil separating primary and secondary coils. I also used plastic insulator to hold both coils away from the core. This was with a resonant circuit and with primary side amplitude regulation and after a 100 uH common mode choke on the secondary side. That circuit gets tricky if you want to reach 1uA rms or less.

Regards, Dieter

[fcb]

Interesting subject. We've used a hugely oversized E3 transformer running at 100KHz in one of products to lower capacitance between two circuits.

I liked the Cleverscope arrangement, but my favourite is the Keithley DMM7510 teardown Dave did that uses coax windings.  Newtons 4th (voltech) have an interesting method using single turns on ferrite rings with copper tape guards.

To get round capacitively coupled noise on something i'm finishing at the moment I've taken to shutting off the PSU, taking a measurement (80us at 30Hz), and then restarting the supply (100KHz).

[fcb]

This I've seen done with a single winding (like the joined secondaries) and with a copper tape screen on the primary.

I'd probably do it with a 3D printed bracket to guide the windings off the core (reduce capacitance further).  This is also used for data transmission in some power-analysers.

[dietert1]

Yes, a similar setup was also used for data transmission in the isolation interface of the HP 3456A DVM. Don't know whether they use that interface during measurement or autozero. I think if the core is kept off by plastic insulator or spacer this will have more capacitive coupling than with (partial) air spacing as in the Cleverscope image. But that's something to decide by measurements.
Up to now i decided to make something old school that is cheap and hard to beat. Anyway, i can't imagine running that mains transformer close to a voltage reference. Neither can i imagine running a high frequency converter near a voltage reference. Integrating and optimizing such a setup will be a major engineering effort.

Regards, Dieter

[doktor pyta]

AFAIK one of the Fluke/ Hart Scientific thermometers used two akkus or super caps. One was charged and the other was connected to the load at the time.
This was patented solution.

[Echo88]

Links to the circuit in the F1595a mentioned by doktor pyta:

https://download.flukecal.com/pub/literature/Recent%20Advances%20in%20Resistance%20Thermometry%20Readouts.pdf
https://patentimages.storage.googleapis.com/94/8a/8b/fdbc915129e6b9/US8120327.pdf

[fcb]

AFAIK one of the Fluke/ Hart Scientific thermometers used two akkus or super caps. One was charged and the other was connected to the load at the time.
This was patented solution.

I quite like that approach.  Could probably work if you can get a high-enough isolation/low enough capacitance between across a relay.

I thought about using a motor, long plastic shaft and generator (in reality two stepper motors), if driven by low frequency could be low noise.  Could even mount motors onto a PCB and use a belt.  Risk of creating a van-der-graff generator?  Although you could run the belt past a numer of generators if you needed to power a few channels.

[The Soulman]

AFAIK one of the Fluke/ Hart Scientific thermometers used two akkus or super caps. One was charged and the other was connected to the load at the time.
This was patented solution.

I quite like that approach.  Could probably work if you can get a high-enough isolation/low enough capacitance between across a relay.

I thought about using a motor, long plastic shaft and generator (in reality two stepper motors), if driven by low frequency could be low noise.  Could even mount motors onto a PCB and use a belt.  Risk of creating a van-der-graff generator?  Although you could run the belt past a numer of generators if you needed to power a few channels.

If you want to go mechanical, why not use two (brushless) electric motors with a belt or a plastic axle between them, one driving the other?

The switched battery/supercap idea is kinda clever tho, something like a beefed up ltc1043.

[fcb]

AFAIK one of the Fluke/ Hart Scientific thermometers used two akkus or super caps. One was charged and the other was connected to the load at the time.
This was patented solution.

I quite like that approach.  Could probably work if you can get a high-enough isolation/low enough capacitance between across a relay.

I thought about using a motor, long plastic shaft and generator (in reality two stepper motors), if driven by low frequency could be low noise.  Could even mount motors onto a PCB and use a belt.  Risk of creating a van-der-graff generator?  Although you could run the belt past a numer of generators if you needed to power a few channels.

If you want to go mechanical, why not use two (brushless) electric motors with a belt or a plastic axle between them, one driving the other?

The switched battery/supercap idea is kinda clever tho, something like a beefed up ltc1043.

I literally said that and you've quoted it.

[The Soulman]

AFAIK one of the Fluke/ Hart Scientific thermometers used two akkus or super caps. One was charged and the other was connected to the load at the time.
This was patented solution.

I quite like that approach.  Could probably work if you can get a high-enough isolation/low enough capacitance between across a relay.

I thought about using a motor, long plastic shaft and generator (in reality two stepper motors), if driven by low frequency could be low noise.  Could even mount motors onto a PCB and use a belt.  Risk of creating a van-der-graff generator?  Although you could run the belt past a numer of generators if you needed to power a few channels.

If you want to go mechanical, why not use two (brushless) electric motors with a belt or a plastic axle between them, one driving the other?

The switched battery/supercap idea is kinda clever tho, something like a beefed up ltc1043.

I literally said that and you've quoted it.

   Sorry about that, first time I've read that to quickly and thought you were thinking of a motor driven commutator to
use instead of a relay.
Reading again, I don't know what I was thinking.

[guenthert]

     I don't quite see the need for extreme measures for the power supply of a DC voltage reference as typically they just feed long scale DMMs measuring DC, which take their sweet time integrating the signal (Fluke 8508A takes a whopping 25s for the conversion at best resolution -- twice as fast as the old Solartron 7081) or being compared to other sources using a null meter (Fluke 845A specifies a patience-testing settling time of 5s at the most sensitive range).  Substituting a nano voltmeter like Keysight's 34420A for the null meter speeds up the measurement considerably, still for best results it'll integrate over multiple power line cycles.  That of course assumes that there won't be some rectifying of the signal in the meter.  For null meters AC rejection is specified and can be tested. 

     I'd think other instruments, e.g. spectrum analyzers, have much more stringent requirements for the noise-lessness of their power supplies.  How are they doing it?

[SilverSolder]

[...]
     I'd think other instruments, e.g. spectrum analyzers, have much more stringent requirements for the noise-lessness of their power supplies.  How are they doing it?

One thing I've seen done is the use of a shielded transformer for the mains supply, to avoid common mode noise capacitively leaking across from primary to secondary, I guess.

[exe]

     I'd think other instruments, e.g. spectrum analyzers, have much more stringent requirements for the noise-lessness of their power supplies.  How are they doing it?

Looking at tear-downs, there are many-many individual power rails in modern instruments, and most if not all of them are smps. I think in R&S instruments I've seen completely integrated modules that don't require any external components. Probably, this way parasitic inductances are small, thus reducing the noise.

I've also seen decoupling capacitor networks, where multiple caps are in parallel. But you can't just wack in several caps and hope they they will have a combined effect. Usually it only makes things worse. So it takes some engineering to do it right.

[Dr. Frank]

     I don't quite see the need for extreme measures for the power supply of a DC voltage reference as typically they just feed long scale DMMs measuring DC, which take their sweet time integrating the signal (Fluke 8508A takes a whopping 25s for the conversion at best resolution -- twice as fast as the old Solartron 7081) or being compared to other sources using a null meter (Fluke 845A specifies a patience-testing settling time of 5s at the most sensitive range).  Substituting a nano voltmeter like Keysight's 34420A for the null meter speeds up the measurement considerably, still for best results it'll integrate over multiple power line cycles.  That of course assumes that there won't be some rectifying of the signal in the meter.  For null meters AC rejection is specified and can be tested. 

Hi guenthert,

I thought just like you, until I found out in practice, that ordinary linear power supplies, w/o shield and isolation, obviously  create ground loops which may shift the reference voltage by several tenths of ppm. These may also manifest in disturbance from mains into the circuit, which may create additional shifts.

The worst effect was observed during the difference measurement between two mains supplied LTZ circuits, using a DMM like 3458A or 34465A, I don't remember. That measurement was complete toast, only huge spikes.

I did not repeat that measurement any more by using my KEI 182A nV meter, but maybe I'll do that now that I have a Fluke 7000, against a mains or battery supplied LTZ. 
Frank

[branadic]

At least a yesterdays measurement with K3458A in 100mV range worked, measuring one of the F7000 (cell2) powered from mains against a F731B, also powered from mains. The readings were stable and consistant. So it is possible, at least on commercial products.

-branadic-

[niner_007]

I quite like that approach.  Could probably work if you can get a high-enough isolation/low enough capacitance between across a relay.

There are a lot of relay options out there, series connected relays might also be an option

[Kleinstein]

Ideally the reference should not be very sensitive to mains hum an similar. In most cases the circuit where such a reference is used / needed is relatively low impedance and thus should not be effected very much. It is more like that most meters would also inject some signal. In some cases good isolation at the reference could compensate for not so ideal isolation at the meter. The main critical cases I see would be using a KVD or Hammon type divider, as these have relatively high impedance.

So chances are one could get away with a relatively simple mains supply (like classical mains transformer with separated windings and linear regulator to avoid possible surprises from EMI). For the few really critical measurements one could use battery power - this is kind of the standard to compare too.

[Dr. Frank]

A few years ago, I made measurements with 3 different setups, always using a 3458A, NPLC 50 in difference or absolute configuration. I used 3 references, Ref_1 and Ref_2 in one case on same supply, LTZ #5 in separate case and with separate supply:

1. difference between two separate references, with 2 separate mains supplies, about 9mV apart
2. difference between two references in one case, on same mains supply, about 47mV apart
3. absolute measurement on one of the references

For all measurements I then calculated the relative deviation in ppm.
First case made really ugly spikes, up to 100 times bigger than the absolute measurement.
Even the 'quiet' areas between 7h-8h were much noisier compared to the absolute measurement.

LTZ #5 meanwhile runs on batteries, so I might repeat this test.

REF_1 and REF_2 were my 16 years old prototypes, standard LT circuit, LTZ #5 is Andreas design, practically immune to external disturbances.
Absolute measurements on latter references, as well as on the FLUKE 7000 usually show noise of max 0.2ppm_pp only, in 'quiet' areas even half of that

Frank

[dietert1]

Here is todays 24h log of two LTFLU references, difference measurement with HP 3456A. Vertical scale is +/- 0.5 ppm again. Those two references are running continuously on mains, using the old school power supplies as shown above. That setup includes a second HP 3456A to measure temperatures, two unmodified Arroyo temperature controllers (with SMPS) and a HP 59401A isolated GPIB DAC for temperature control. This represents development status of May 2020.

Regards, Dieter

[MK]

The construction details of the GR shielded bridge transformers are interesting- https://www.ietlabs.com/pdf/GR_Experimenters/1935/GenRad_Experimenter_Oct_1935.pdf
You can also shield a toroid with copper tape.
Me, I'd probably use a couple solar cells, lit up with some big incandescent lamp! Can't get much better isolation than that.

I have a couple of double shielded tranformers made by Singer/Gertsch and they are also 1:4.

What is the reason for these double-shielded transformers often being made with a 1:4 turns ratio?

Perhaps that guarantees that the output volts count as "SELV" even with 230V going in?

[MiDi]

Perhaps that guarantees that the output volts count as "SELV" even with 230V going in?

SELV has to be <= 50 VAC or 120 VDC in air and w/o protective earth connection.

[branadic]

Xmas time, reverse engineering time.
LCR meter powered up to measure some inductance, multimeters to measure coil resistance, caliper to get some dimensional measurements and handheld multimeter to measure frequency on the F7000 with LT1533.
Based on that I will try to replicate this mystical Pickering xformer circuitry for some DIY voltage references I have. A spice circuit was already build to prove the reverse engineering was correct.

-branadic-

[dietert1]

What is the coupling capacitance from primary to secondary of the transformer?

Regards, Dieter

[branadic]

Forgot to measure it and when I thought of it, the unit was already put back together. But I can measure it once my replica is done. Boards already designed, need to order them in the next step.

-branadic-

[Andreas]

Hm,

in the paper I read from 2 "shield screens" (primary + secondary) consisting out ouf a polymer.
Where are the shields connected to? Case or Guard?

With best regards

Andreas

[Echo88]

Measured the coupling capacitance of two DIY-Pickering Transformer without shield:

13pF and 21pF

Other normal COTS transformers suitable for the LT1533 have the following coupling capacitance for comparison:

CTX02-13834 (suggested in the LT1533 datasheet): 9pF
WE 314619 (substitute for some other CTX02): 25pF
WE 30804 (substitute for some other CTX02): 86pF

Measured with HP4192A at 1MHz.

In the end the ultimate test is to see wether someone can spot the difference when the powersupply is switched over from battery to DCDC, while measuring the 10V reference in opposition-mode.

[dietert1]

I'd guess the one shown above by fcb will be like 1 pF.
There is nothing mystical about that pickering transformer. The two core idea is the same, except the position of the two cores is different, since it is meant for power transfer. So you don't want one but more than one turn for the coupling coil.
From the images i can see that the DIY Pickering transformers have more distance between the cores than the original one. So the original one will have more coupling capacitance, maybe 30 or 40 pF. And the shields don't really help, since the guard can't absorb high frequency spikes very well.

Regards, Dieter

[branadic]

Andreas, primary shield is tied to primary GND and earth / case, secondary shield is tied to secondary GND.

-branadic-

[Kleinstein]

I have build an ugly DC/DC converter based on a relatively large core with distance to the windings. Still no shield, but the coupling capacitance looks small, at some 3-7 pF  (hard to measure) range.

[Andreas]

Andreas, primary shield is tied to primary GND and earth / case,

Earth / housing is that what I would have expected.
But obviously the connection is not short (< some mm) and routed via GND.
The transformer position on a PCB is not a optimum. It should be tied directly to case.

secondary shield is tied to secondary GND.

I would have expected to a guard (inner housing)
Secondary GND does not make any sense for me. (will not reduce coupling capacity).

with best regards

Andreas

[fcb]

How are you measuring the capacitance? LCR meter?

One way I've done it is to measure the leakage current increase at 3200V 50Hz. Seems to correlate quite well with an LCR meter when measuring 10-20pF. And we seem to be able to measure reliably down to 1pF with the 3200V/50Hz - which the LCR meter seems to struggle with (the only 4 terminal LCR meter we have was homebrew).

Maxim (MAX14436) make some nice couplers that are around 1pF per channel BTW.

[Echo88]

https://eu.mouser.com/datasheet/2/427/vom617a-1767067.pdf 0.3pF and no EMI from the Digital Isolators
There might be even better Optocoupler, i just found it after a bit of searching.

[fcb]

https://eu.mouser.com/datasheet/2/427/vom617a-1767067.pdf 0.3pF and no EMI from the Digital Isolators
There might be even better Optocoupler, i just found it after a bit of searching.

VOM617, great device, use on numerous projects.  But it can't do 200Mbps

[Echo88]

Sorry, was thinking about power supplies and feedback-coupling and not about your data transmission project.

[branadic]

Earth / housing is that what I would have expected.
But obviously the connection is not short (< some mm) and routed via GND.
The transformer position on a PCB is not a optimum. It should be tied directly to case.

If you carefully look at the first picture you see a nut between primary and secondary side of the transformer, to which the case is connected to.

I would have expected to a guard (inner housing)

There is neither a guard nor an inner housing in this units.

Board design finished and order placed.

-branadic-

[branadic]

Received some ring cores before new year and started to wind some copper on them. Since we still have lockdown and even the shops where you can buy some enameled wire are closed, I had to use what I had in the drawer, thus I used some teflon wire.
The result I found is, that 2x 16 (17) turns on the primary side (~4.2mH per coil) and 2x 32 turns on the secondary side (~14mH) do match the inductance I measured on F7000 with LT1533, thus a ratio of 1:2. The transformer is wound in a bifilar fassion. Not sure if twisting the wires would make any difference though.

The boards I designed already arrived in Germany and will hopefully show up soon. It will take some additional time to draw the resistive shells in CAD and have them printed. Not yet sure if my idea works, but stay tuned.

-branadic-

[Kleinstein]

I think it should be OK with these cores (low magnetostriction), but with soft ferrite cores one should keep in mind the effect of mechanical stress on the core.  It can reduce the Al value a lot - those ferrite cores can make sensitive force sensors.

For the coupling coil it can help to have quite some distance. Probably no need for that many coupling "turns / loops": 3 - 5 may be sufficient.  Multiple separate loops may be better than more turns in series. There is still the one "turn" in parallel to the core that could upset things.

[robert.rozee]

doktor pyta (reply #3) and Kleinstein (reply #50):

i'm picking you're both using a driver circuit similar to this:


https://en.wikipedia.org/wiki/Royer_oscillator#/media/File:Royer_oscillator.gif

are you able to post a schematic to what you are using with some component values? i'm aiming a fair bit lower myself, just after something to go from a 6v battery up to 20v with relatively little noise to run a LM399. i'm led to believe that my present solution - an LM2577 boost converter - is frowned upon in present company!


cheers,
rob   :-)

[doktor pyta]

@robert.rozee

kind of, but I'm using 2xNMOS.

In other project I've successfully implemented circuit described in article below (but with center tapped primary coil).
This is a very good article worth reading.

https://www.mdpi.com/1996-1073/11/10/2653

[Kleinstein]

My converter uses an external drive signal for the transistors, so not self resonant, but a fixed drive push pull transformer. Different from common implementations there is a relatively long dead time and quite some RC damping.

[Gyro]

I hesitate to post this as it's a completely off the wall idea. I've accumulated a couple of old Oral B style electric toothbrushes over the years which have small magnetic charging bases  - the ones with the little plastic spigot that goes into the base of the toothbrush.

I haven't done any tests yet, but inside the toothbrush, there is just a simple coil on quite a large diameter standard former. No ferrites. It occurs to me that this has the basis of a really compact mains powered, high isolation, low capacitance PSU. The large gap and full mains insulation of the charger base would make it really easy to apply grounded screening foil (obviously no shorted turn). There's enough energy transfer to charge an AA sized NiMh in around 8 hours, so it ought to be sufficient for a low current reference with a coil rewind.

Obviously this is a one-off type solution, and maybe the switching waveform is too noisy, but it might just trigger some ideas.

[macaba]

I hesitate to post this as it's a completely off the wall idea.

As it happens, I had the same toothbrush in my disposal pile!

So some quick approximate measurements with DMM6500 in Digitize Current mode measuring from one side of the battery inside the toothbrush to PE on the back of the DMM6500.

DMM6500 noise floor: 0.005uApp
Without charge base nearby/on: 0.4uApp
Without charge base nearby/on, inside thick steel box section grounded to PE through 1M resistor: 0.2uApp
Whilst on charge base: 1.5uApp

As is, no good; the modifications you suggested would be a good start but not something I'm interested in doing at the moment.

Edit: I'm unsure if there might be some interaction with the DMM6500's own leakage (0.14uApp)

[exe]

maybe the switching waveform is too noisy, but it might just trigger some ideas.

https://youtu.be/JJgKfTW53uo?t=1320 . This, of course, rerpresents only one model from one brand. Other may do it better.

[Gyro]

I hesitate to post this as it's a completely off the wall idea.

As it happens, I had the same toothbrush in my disposal pile!

So some quick approximate measurements with DMM6500 in Digitize Current mode measuring from one side of the battery inside the toothbrush to PE on the back of the DMM6500.

DMM6500 noise floor: 0.005uApp
Without charge base nearby/on: 0.4uApp
Without charge base nearby/on, inside thick steel box section grounded to PE through 1M resistor: 0.2uApp
Whilst on charge base: 1.5uApp

As is, no good; the modifications you suggested would be a good start but not something I'm interested in doing at the moment.

Edit: I'm unsure if there might be some interaction with the DMM6500's own leakage (0.14uApp)

Wow, some impressively quick measurements, thanks!

As you say, it sounds like no good as it stands, but scope for experimentation. There's certainly enough physical gap, without the toothbrush body to implement some sort of primary screen, and a separate guard screen on the secondary.

https://youtu.be/JJgKfTW53uo?t=1320 . This, of course, rerpresents only one model from one brand. Other may do it better.

Thanks exe, I hadn't realised that Dave had done a teardown, well of the toothbrush side, I think all of the charging bases are potted internally for waterproofing. The one that I've pulled apart is a single winding with a bridge rectifier rather than a lower loss 2 winding two diode full wave arrangement.

Dave showed some interesting waveforms, interesting about the 800Hz charge pulsing - I had assumed that they were 'dumb' HF oscillators but maybe there's some (unreachable) intelligence in there.

I will report back any findings.

[Kleinstein]

The charging intelligence is at the receiver side, not at the driver side. The receiver side would be rebuild and should avoid too much of a variable load.

A main problem may be electric fields from the drive side, as the coils are likely high voltage. So a primary shield can do a lot.

With a not so well closed magnetic field, there can also be some trouble from the stray magnetic field, not just the electric field. So it may need some distance to the circuit.

[dietert1]

...
Edit: I'm unsure if there might be some interaction with the DMM6500's own leakage (0.14uApp)

That is a very low number and it made me look through the DMM6500 teardown on xdevs.com. There is a double mains transformer, both cores with two chambers each and with shield. There could be an additional high frequency converter hidden under an aluminum cover.

In my understanding one critical test case are the kilovolt pulses arising from mains switches of inductive loads. That cannot be solved by a low coupling capacitance alone, because coupling below some pF is unrealistic. So there should be a shield connected to the guard net. A 12VA commercial shielded mains transformer i measured recently had coupling from primary to core+shield of about 260 pF (!) and a similar coupling between secondary and shield. With that capacitance even a 15 Vac secondary at 50 Hz injects several uA into Gnd. Definitely not what you want, although the coupling around the shield is only 2.7 pF.

Regards, Dieter

[TexasRanger]

Why not use a little hydro generator - Oil - cooling pump combination?

https://de.aliexpress.com/item/33009775552.html?spm=a2g0o.productlist.0.0.7eda53a6MhrDSy&algo_pvid=d646b410-b034-47a4-8a78-a6cd88366715&algo_expid=d646b410-b034-47a4-8a78-a6cd88366715-0&btsid=0b0a0ac216097101883665950e81bd&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

Would for sure be a funny project and would achieve femto range parasitic capacitance and MV proof insulation.

[Kleinstein]

The point of he shield is to reduce the voltage to an absolute minimum, like less than the voltage per turn - ideally a small fraction of this if the point were the shield is connected is chosen right. With a small mains transformer it is something like 1/10 V per turn. If one has 15 V at the shield there is something wrong.
260 pF coupling sounds relatively high, but with a low voltage this may still be acceptable. It only needs a good ground to connect too.

The capacity from primary to shield is usually not important. The point is the capacitance between the 2 shields.

In the DMM6500 only the transformer for the analog part needs to be shielded. The digital/output  part may be even grounded - so no real need for a special transformer for that part.  I doubt there would be another switched mode converter - at least in the analog part.

[dietert1]

Maybe i need to explain better.
You want to connect the shield of the shielded transformer to your guard network to keep away noise coming in from primary (mains). Then you build some circuit (rectifier whatsoever) that makes a connection between the secondary and Gnd/Low of the device you want to supply. Now you will have AC current along the capacitance between shield and secondary injected between Guard and Gnd. This current flows through the Gnd/Low network to the one point were you wanted to connect Guard and Gnd/Low originally. On its way it generates error voltages in the Gnd/Low network. When i calculate that current at 15 Vac 50 Hz 260 pF i get 3.5 uApp. Compare this to the number 0.14 uApp given above for the DMM6500.

The water generator proposal is interesting. If those devices work well with oil, one could combine two of the same, one as pump the other as generator, except it will make some noise and last "only" about 3000 hours. That's what they claim. Maybe it can last longer with the low power required for a voltage reference.

Regards, Dieter

[robert.rozee]

@robert.rozee
kind of, but I'm using 2xNMOS.
In other project I've successfully implemented circuit described in article below (but with center tapped primary coil).
This is a very good article worth reading.
https://www.mdpi.com/1996-1073/11/10/2653

the article is interesting, but a tad more theoretical than i was hoping for.

however, this posting:

https://www.eevblog.com/forum/projects/inductive-charging-how/msg139940/#msg139940

points to an actual implementation that someone has built here:

https://www.mikrocontroller.net/articles/Royer_Converter
https://translate.google.com/translate?hl=en&sl=de&u=https://www.mikrocontroller.net/articles/Royer_Converter&prev=search&pto=aue (translated)



for my own requirements (no isolation, 6v battery) i'm thinking i probably don't even need the secondary winding - if the primary presents a clean enough sinewave, this can just feed a pair of cascade voltage multipliers to get a clean DC output for the reference (20v at probably less than 30mA).


cheers,
rob   :-)

[Echo88]

Remember to choose the right oil and test the tightness of the oil drain plugs on your voltage references.
Just like in those Fluke 720A. 

btw:
Attached is a short simulation of the supercap-switching-circuit like used in the Fluke 1595A discussed a few threads ago, to get general ballpark numbers.
If relays like the AGQ200 (we want very low capacitive coupling) would be used with its lifetime of minimum 100000 cycles at nominal electric load, then we could expect a lifetime of the relay in this simulation of at least 3 years. Stick them into sockets and replace every 3 years as a maintenance-cycle...certainly easier than replacing the oil-pumps 

https://www.buerklin.com/medias/sys_master/root/h5c/hea/9313902428190/8873768026142.pdf

[exe]

The next thread would be what type of oil is the best for volt-nutting.

[Gyro]

I think that one's already been done.     https://www.eevblog.com/forum/metrology/oil-for-a-leaking-tinsley-standard-resistor/

[antintedo]

If relays like the AGQ200 (we want very low capacitive coupling)

Pickering 103 series might be more suitable - they have guard shield and the capacitance is specified in the datasheet, it's on the order of 0.1pF. They have much higher lifetime as well. Some small signal relays have high frequency isolation specs. I've seen this in G6K series, capacitance derived from the graph was <0.2pF.

[Echo88]

Unfortunately the Pickerings arent easily available from Mouser/Digikey. But the general idea to use Guarded Relays is nice. Can you explain how you derived the coupling capacitance via hf-isolation spec, since im not a hf-guy?

[antintedo]

Can you explain how you derived the coupling capacitance via hf-isolation spec, since im not a hf-guy?

Me neither, but my understanding is the isolation test can be modeled as a voltage divider with a capacitor in one leg and 50 ohm termination in the other. Capacitor impedance will be 1/(2pi*f*c), so substituting numbers, for G6K -86dB at 1MHz the capacitance turns out to be 0.16pF.

[dietert1]

A small reed contact from my drawer wasn't more than 0.1 pF (a bit difficult to measure). Another, larger one had less contact distance and it gave about 1 pF. Then with a reed relay there will be additional coupling along the coil, so to keep coupling low a metal foil gets inserted between contact and coil. The Pickering relay has that shield and researching the web i saw others like that. I also measured an Aromat RF relay in a VXI E1473A multiplexer, they are below 0.1 pF as well.
The same was true for a Teledyne 732-12, once i connected Guard to the metal can. That one is DPDT, so it replaces four of the Pickerings. But it's small and the wires close and difficult to guard from each other.

General purpose DIL type relays in plastic enclosure had about 0.4 pF. There are special RF relays like Axicom HF3 with 60 dB isolation at 2 GHz. -60 dB in a 50 Ohm system indicates an isolation impedance of 50 KOhm at 2 GHz. That is a capacitor of 0.0016 pF. Strange. Have seen some RF relays where they create shorts on the inactive ports to improve isolation.

If this discussion is about building a multiplexer for two rechargeable batteries, then there will be eight contacts, with four of them open at a time. Then you put the batteries into two separate guarded compartments and you may want to switch those guards between primary and secondary, so maybe five open contacts at a time. Plus there will be stray capacitance in the wiring near the relays. If you get down to about 1 or 2 pF coupling in the end that would be a nice result, similar to the DMM6500 number given above.

Regards, Dieter

[antintedo]

Maybe this image and an old version of the datasheet will shed some light on the construction of HF3 relay:
http://mt-system.ru/sites/default/files/docs/documents/File/TYCO_PDF/HF3.pdf

It might be that there is no direct line of sight between common and switchable contacts. There are some recessed spaces in the base of the relay where signal contacts are supposed to be. When the armature is not connected, it might be shorted to guard - big yellow contact on the left of the image. I could not find any relevant patents, so the best course of action would be to buy one and disassemble it.

[Kleinstein]

The relay looks very much like using 2 contacts in series and maybe even connecting he movable part to ground when open. This can explain very good RF isolation. However the good RF isolation does not mean there is low capacitance of the additional ground (would be an extra shield potential) is not there.

Even when not using the extra shield / ground it can still be a good relais.

[branadic]

Some progress to share:
- boards designed, ordered and received
- cores for primary and secondary side wound
- CAD design of the shield and arrangement drawn
- first samples of the shield fabricated and checked for fitting into each other
- finalized the CAD drawings (dimensions slightly adjusted to tighter tolerances)

Now the final parts are being fabricated. Components on order, so hopefully a first version can be assembled and tested soon.

-branadic-

[EmmanuelFaure]

Nice job branadic! Is the plastic shield made of some conductive plastic for EMI shielding?

[branadic]

Exactly, the shield is made from a conductive material.
Meanwhile I got all parts done to complete at least the transformer.

With finally installed transformer cables to the board, but rest of the PCB unpopulated, I can measure some 17 - 18pF between primary and secondary ground. However, on the original 7000 board some 1nF/1kV cap (blue) can be seen. So not yet sure what this number means by itself.

-branadic-

[dietert1]

In my opinion that 1 nF capacitor is a mistake unless it is necessary to protect the optocoupler from malfunction. Are you using an optocoupler with built in guard?

Like Andreas proposed above, with two shields of the transformer you have several options how to wire them. When making a guarded solution you can connect the coupling wire to guard - it doesn't have voltage on it. Then the optocoupler should move next to the transformer and maybe get a guard trace as well.

Regards, Dieter

[fcb]

17-18pF seems quite high, is the complexity worth it? I can see that the isolation voltage would be reasonable and noise shielding possible.

[Kleinstein]

17-18 pF looks quite high. This should be capacitance between 2 shields - though I somewhat don't see the connection in the pictures.
So the injected current can still be small, as there should be essentially no AC voltage across the capacitor.
The shown transformer could be relatively high power - from the sizes I would consider 20-100 W possible.

If needed one could probably reduce the capacitance with a little more space between the coupling windings and the inner part. For lower  power thinner and less (e..g. 4-5 loops at 0.6 mm²) wires should be Ok too.

[branadic]

Just to clarify, the transformer shown above is a direct replica of the Pickering in F7000 with LT1533, so same dimensions, same amount of copper etc.
With respect to the report linked in the first post, this circuit and transformer was designed for 7W.
The capacitance was measured between primary GND/shield and secondary GND/shield.

-branadic-

[ignilux]

Hi, all-

I just want to clarify my understanding of the concern over transformer capacitance. The principal issue seems to be the coupling of noise between the outside world (primary side) and the circuit (secondary side), correct? Magnetic coupling is largely mitigated with a toroidal core, but that does nothing for electrical coupling through the parasitic capacitance. Hence, trying to minimize that quantity. Am I all good so far?

Here's where it gets a bit fuzzier for me. On first glance, I would assume that most of the electrically coupled noise is common mode, since the noise would have coupled equally to both the hot and neutral conductor in unshielded mains wiring (Romex, in the US). If that's the case, why couldn't the noise be reduced with the use of a common mode choke on the primary and/or secondary side of the transformer? There must be more to it than that, otherwise all of the knowledgeable people here would have suggested it by now.

Anyway, I'm in the process of building a power supply for my AD587JQ and was thinking of starting a thread on the subject, so I'm glad to have found this discussion. Branadic, what type of toroid core are you using in your DIY Pickering? Sorry if I missed that.

Cheers!

[Kleinstein]

The main problem with a simple transformer is common mode injection due to point with different voltage in the coils.  The shield can reduce the voltage to less than the votlage of one turn - below that one has to balance and chose the grounding point right.  So even with a shield, low capacitoance is good.
A common mode choke is of limited use, when at a very good isolation level. It is only am additional seires impedance and this may not be much compared to some 20 pF of capacitance at the transformer.
It mainly helps to the the higher frequency part. For the lower frequency part a low capacitance transformer is more effective.
I would still consider a CM choke worth while, but the transformer should still be low capacitance, especially if the shield is not good.

There may be also the case that the rest of the circuit has some CM voltage to ground - so good isolation is definitely helping and even just capacitance to ground can be negative.
 

[Andreas]

If that's the case, why couldn't the noise be reduced with the use of a common mode choke on the primary and/or secondary side of the transformer? There must be more to it than that, otherwise all of the knowledgeable people here would have suggested it by now.

Hello,

the frequency ranges where a shielded transformer (< 1 MHz) and a common mode choke (> 1MHz) work effectively is different.
So usually you have to use both.

Anyway, I'm in the process of building a power supply for my AD587JQ and was thinking of starting a thread on the subject, so I'm glad to have found this discussion.

Mhm,

for a AD587 reference (< 5mA) I would always use batteries. 12 AAA cells last 7-10 days before you need to recharge.
See also my AD587LW design:

https://www.eevblog.com/forum/metrology/ad587lw-10v-precision-travel-standard/msg1449494/#msg1449494

with best regards

Andreas

[dietert1]

Yes, a common mode choke helps, too. And it isn't easy to make one with low parasitic capacitance, either. Sometimes people combine common mode chokes with coils and sheath current filter (ferrite on cable). A primary side common mode choke works best when combined with a guard. Then you get an inductive divider between the choke inductance and the effective inductance of the guard net.
As always some people try to avoid own studies by "best practices" approach, others try to get own hands-on experience. When i made a short on the probe tip (with its own ground clip), put the probe onto the table without any other connection, adjusted the trigger for mV and turned on/off the microscope illumination (transformer inside!), i captured how that mains switch sparked. For precision measurements you don't want to have spikes of several mV or tens of mV in your circuit, so it will have to be shielded/guarded, like "best practice" tells you for any other audio or RF setup. Some people even connect an amplifier/headphone to critical signals and listen to them..

Regards, Dieter

[ignilux]

Thanks for the replies. I hadn't really considered the orders of magnitude difference between frequencies where each solution is effective, but that definitely makes sense.

Andreas, I'm aware of that thread and am basing much of the project on your design. For the time being my most precise voltmeter is a 5.5 digit Fluke 8840A, so I don't feel the need to go all the way with ppm tempco compensation, etc. It's more for the learning process, and to have a accurate reference on hand for other uses. To that end, I'd prefer to have a mains-powered device that I can leave powered on 24/7. Sealed lead acid might be nice, too, if I go for a battery powered solution. That would extend the run time quite a bit, eh?

Anyway, thanks again!

[KT88]

@Dietert1: Common-mode chokes won't help if one side is floating as a filter consits of a passing- and a shunting element. The shunting element is missing here. There would be some capacitance against ambient though - this would unfortunately resonate at some frequency...
The most viable measures to mitigate CM noise in this case is reducition of coupling capacitance, shielding and symmetry.

Cheers

Andreas

[dietert1]

No, sorry, a primary side common mode filter is an established method to keep the guard/shield clean.

As i wrote, a common mode choke helps in combination with guard. In that case it reduces excitation of the guard net by spikes from outside. Ideally guard would be a Faraday cage, with a very high resonance frequency depending on its dimensions. Below that frequency the inner of the cage would be well protected against excitation from outside. But in general it isn't. It consists of many partial guards in many instruments, connected to each other by (inductive) wiring. So the resonant frequency of the guard net will be somewhat lower. By the way introducing some 50 to 100 Ohm resistors into the guard net allows you to make measurements.

Regards, Dieter

[KT88]

I agree on the placement at the primary side. I was referring to your first, more genral statement. I have seen a lot of examples where CMCs were placed on floating parts of a setup without the desired result of reducing EMI. I some cases during EMC tests some frequencies were even exaggerated depending on cable length...

Cheers

Andreas

[d-smes]

Exactly, the shield is made from a conductive material.

If you used conductive plastic, beware that this creates a shorted turn on the torroid.  I suspect resistivity of the conductive plastic is high enough, or that contact resistance between the two halves is high enough, that the current in the shorted turn becomes meaningless.

[Echo88]

@d-smes: The shields are designed to be just conductive enough to fulfill the shielding while only dissipating neglegible power.
Please read https://eprintspublications.npl.co.uk/1888/1/bemc99-7.pdf page 3 Figure 6 discription.

[exe]

@branadic what plastic did you use for the shield?

[ignilux]

What's the consensus on so-called "semi-toroidal" transformers like this series from Bel Signal? It looks to me like an R-core, but made with a laminated core rather than a secret sauce composite.

[dietert1]

Have seen those in a Keithley 213 quad isolated DAC. They have low coupling capacitance as they are. And the advantage that you have access to the core. I remember bending up a corner of the uppermost core sheet and soldering a wire to it to connect the core to guard.

Regards, Dieter

[The Soulman]

Any drift for both the meter and the ltz1000 should be down in the noise for such a short time and temperature range.
Did you let everything warm-up (for several hours) before starting the measurement/data-logging?

[dietert1]

Your reference output voltage increased by about 1 mV. In your own words this is bad. To get anywhere near the expected results, one would use voltage regulators between batteries and reference board. As far as i remember that HP 3458A reference board needs bipolar supply to work properly.
Also i was wondering, whether that BNC cable on your black box is the reference voltage output.

Regards, Dieter

[TiN]

Something is awfully wrong. Ditch the LiPo (it's there just waiting you to short something by accident and blow stuff up) and use 36313A PSU to test first.
At first glance you seem to forgot connect heater supply or returns to the PSU, as that could be nearly only possible explanation to this insane instability. Tempco of most A9 boards is beyond what any of the 6.5d DMMs can measure in small temperature swings (<10°C).

Why you don't show more photos showing exact setup , connections, wiring etc? It's not year 2005 when you had to pay $$ for every megabyte of internet data on dialup  .

[Kleinstein]

The ground to the heater power must be connected to the main ground - just keep the output separate. It is just inside the DMM circuit that heater ground and power ground use separate paths, but they are still connected on the PCB.

The negative voltage is only used to reduce the ground return current - this is not so much an issue here. For testing only the reference, one should get away without using it: one mainly misssed the extra heat source.
It would help to have a regulated voltage for the main part. The heater may use the same battery pack with a separate regulator or even before the regulator.

[dietert1]

Heater current 0.045 mA means the oven is inactive. The LTZ1000 datasheet recommends 0.1 to 0.3 W heater power. Depending on heater voltage it should take at least 10 mA or so.
And the schematic in the image shows the heater minus separate from the reference Gnd, so in case of using two separate supplies one should make an external connection.

Regards, Dieter

[TiN]

Either LTZ chip is blown open now, or power is not properly supplied. Normal heater currents for A9 are about 25-29mA for heater and few mA for signal ground.

As pointed out, ground MUST be connected as PSU end. Easy to check if heater works by looking at control temperature input (1/13k divider tap output) and control feedback (opamp output for oven control). Both should be same voltage around 0.4** V-ish. You should have four wires going to PSU which are +12...+15V for signal power (opamp, etc), signal ground return, +15 or so power for heater in, heater return to PSU-. Then short circult jumper between returns at PSU LO ports, if your PSU has isolated outputs (like E36313A does).
Then you have your sense lines for zener+ and zener- from A9 to voltmeter.

[MegaVolt]

If you click on the virtual display 6500 there is an option to save only the screen. In my opinion, this is more convenient than saving entire screenshots.

[iMo]

The first thing I would do - I would replace that cheapo LT1013 dil8 socket with precision one. I had many problems decades back with digital stuff not mentioning analog while messing with those sockets.. Unless the precision socket is not suitable there because of different plating..

[branadic]

@ ViewFinder

Would be great if you would either open a new thread or chose another one, as this thread is intended to address power supplies for voltage references and your specific problem has nothing to do with it, as far as I can see. *no hijacking*

@ exe
The plastic I used for the shield is Proto Pasta, the shield itself is 3D-printed

Still waiting for components, while I already build a second transformer.

-branadic-

[dietert1]

I think the engineering approach you propose is completely valid. Maybe the OP is a salesman and his thread title was really meant as "Look at this power supply for voltage references". Then people like you and me make it into a wild collection of personal taste, belief and knowledge. A true mishap.

Regards, Dieter

[Kleinstein]

The demand on the power supply is mainly low injetion of common mode signal, low capacitance and good EMI compatibilty.  It is relatively hard to tel how a reference circuit reacts to EMI - there can be resonances. In addition the reference is not alone, but it would be connected to some meter or similar and these may to be sensitive.

It is still a good question on how to quantify the quality - there are different frequencies involved and the capacitance from the output side  to ground can make a difference. Chances are it would be more the current to a low impedance connection that would matter, not so much the open ciruit voltage, with little extra capacitance. 

[TiN]

Branadic hardly a sales person  . And thread started for PSU design talk and sharing results, not for fixing particular altered and possibly damaged LTZ ref.

Before going EMI rabbit hole it is better to apply requirements expected from voltage reference application. If its going to be measured by floating DVM and null-meters against other references then major required feature of PSU is low leakages and common rejection to avoid ground loops. If reference is used as a part of bigger system (like A9 3458A module) it have different requirements for PSU. Above shown A9 module is very sensitive for pickup and grounding. It is essentially useless as transfer reference on ppm level. Just turning common LED lamp with crappy SMPS will ruin your A9 output stability and noise. No matter if powered by nude virgin custom ppmium transformer or  fire breathing lithium pack. And thats offtopic here, we already have excellent RFI\EMI test thread right here.

I really don't understand View[]Finder's resistance to make own thread named "Experiments with DMMs and references" and post all stuff there, people will be much happier to help and suggest things when it's all in one dedicated spot, instead of guessworking thru 5 different hijacked threads on what is going in. Threads creation does not require payment or anything  PPMs also nothing to do with external data, that is completely wrong, just like saying percents are anything to do with calibrations. It's just a convenient tool to do the opposite - remove absolute values from measurands and outline the deviation of one set with arbitrary selected point(s) against another.

to find a way to have a $200 standard perform as well as a $2,000 one

$200 standard will stay such, no matter what magical power is used to power it. Noise is least of the problems in voltage references, compared to much harder issues like stability, temperature/humidity/pressure dependence, jumps and stress factors, robustness during transport, etc.

[TiN]

Threads are designed as tool to organize things. This one is about power supplies for voltage references. Not about references themselves or how to measure/analyze them as ultimate goal. It is nice and respectable to keep metrology section (and any other section same way) organized and structured, instead of having all threads go in random directions, also to help newcomers to find information easier. Branadic got into power supply design reference by study on modules he got, and its quite interesting direction. Your results with A9 board seem to be all over and so far unique to your setup/equipment/case/device. I can't speak for everyone here, but instead of trying to reply to your posts in every somewhat remotely related threads about this I would be more inclined to spend time helping you rectify issues in separate dedicated thread. From posts it is visible that you misunderstand some key concepts (nothing wrong with that!, we all learning here) so those warrant to be separated from power supply design or repair of my 3458A U2 as well. PPM means parts per million. It does not mean parts per ideal 10V, you don't subtract anything to calculate ppm deviation of value A to B.

others' offers of advice were welcome and I would gladly have accepted a suggestion to relocate that discussion. None was offered.

I've suggested to create your own thread twice in two different threads already. I think you just trolling now.
Why make it hard on people to navigate thru multiple offtopic discussions? To save an entry in forum database?

The reason why EEVBlog is one of the best forums about electronics is variety of topics and all different people who have shared engineering knowledge about those things in all those threads. Trying to keep it structured is least we can do before we drown in huge pile with everything mixed in a single thread. I'm done, sorry all.

[View[+]Finder]

Threads are designed as tool to organize things. This one is about power supplies for voltage references. Not about references themselves or how to measure/analyze them as ultimate goal. It is nice and respectable to keep metrology section (and any other section same way) organized and structured, instead of having all threads go in random directions, also to help newcomers to find information easier. Branadic got into power supply design reference by study on modules he got, and its quite interesting direction. Your results with A9 board seem to be all over and so far unique to your setup/equipment/case/device. I can't speak for everyone here, but instead of trying to reply to your posts in every somewhat remotely related threads about this I would be more inclined to spend time helping you rectify issues in separate dedicated thread. From posts it is visible that you misunderstand some key concepts (nothing wrong with that!, we all learning here) so those warrant to be separated from power supply design or repair of my 3458A U2 as well. PPM means parts per million. It does not mean parts per ideal 10V, you don't subtract anything to calculate ppm deviation of value A to B.

others' offers of advice were welcome and I would gladly have accepted a suggestion to relocate that discussion. None was offered.

I've suggested to create your own thread twice in two different threads already. I think you just trolling now.
Why make it hard on people to navigate thru multiple offtopic discussions? To save an entry in forum database?

The reason why EEVBlog is one of the best forums about electronics is variety of topics and all different people who have shared engineering knowledge about those things in all those threads. Trying to keep it structured is least we can do before we drown in huge pile with everything mixed in a single thread. I'm done, sorry all.

Please accept my apologies for the confusion this has caused--particularly to those who have offered me assistance in the past--and I assure you that I have no intent to "troll" on this or any other forum. I remain interested in contributing to the discussion centers on measurement, analysis and reduction of errors in voltage as measured by the kind of instruments I own: former premium-grade Keithley, Keysight and Fluke products. I believe that is consistent with the interests of the general readership. Of course I will be happy to start another thread. Would it be asking to much to request suggestions for a title and, please, may I place it under the Metrology section?

Best regards,
Donal B. Botkin

[Gyro]

I hesitate to post this as it's a completely off the wall idea.

As it happens, I had the same toothbrush in my disposal pile!

So some quick approximate measurements with DMM6500 in Digitize Current mode measuring from one side of the battery inside the toothbrush to PE on the back of the DMM6500.

DMM6500 noise floor: 0.005uApp
Without charge base nearby/on: 0.4uApp
Without charge base nearby/on, inside thick steel box section grounded to PE through 1M resistor: 0.2uApp
Whilst on charge base: 1.5uApp

As is, no good; the modifications you suggested would be a good start but not something I'm interested in doing at the moment.

Edit: I'm unsure if there might be some interaction with the DMM6500's own leakage (0.14uApp)

Wow, some impressively quick measurements, thanks!

As you say, it sounds like no good as it stands, but scope for experimentation. There's certainly enough physical gap, without the toothbrush body to implement some sort of primary screen, and a separate guard screen on the secondary.

https://youtu.be/JJgKfTW53uo?t=1320 . This, of course, rerpresents only one model from one brand. Other may do it better.

Thanks exe, I hadn't realised that Dave had done a teardown, well of the toothbrush side, I think all of the charging bases are potted internally for waterproofing. The one that I've pulled apart is a single winding with a bridge rectifier rather than a lower loss 2 winding two diode full wave arrangement.

Dave showed some interesting waveforms, interesting about the 800Hz charge pulsing - I had assumed that they were 'dumb' HF oscillators but maybe there's some (unreachable) intelligence in there.

I will report back any findings.

Just a quick follow-up. I just posted a quick teardown of the charging base for a Lidl torch in Dave's original Braun toothbrush teardown discussion thread... https://www.eevblog.com/forum/blog/eevblog-284-braun-toothbrush-teardown/msg3451180/#msg3451180

It seems to have similar power output capability to the Braun but operates at a higher frequency, in the region of 150kHz, and without the severe ringing that Dave observed. Surprisingly it's just a single transistor oscillator circuit. I was expecting something a bit more complex to cope with the inductive load difference between zero load (removed) and full charging load.

Hopefully of some interest for a high isolation supply anyway if it is quieter. The higher operating frequency might be easier (or more difficult!) to filter out.

[dietert1]

Yesterday i found this image taken by ve7xen in a Datron 1081 multimeter. Those guarded transformers look like the solution i arrived at in my study of the "Art of electronics" proposal. Except each of those transformers includes a single turn winding (green wires). How does that work?

Regards, Dieter

[Vgkid]

Those look like isolation transformers for the digital side. They were widely implemented before the use of opto-isolators Here is an example in the HP 3456A.

[Kleinstein]

The circuit is like 2 transformers in series: First some 2x10 to 1 turn and than back 1 to 2x10 turns. One turn helps to reduce the capacitance of a single transformer and the 2nd transformer halves the capacitance and brings the signal back to a level were one wants it.

The transformer with 2 cores close by and one coupling winding is a similar principle, just a few more windings for coupling and less distance. The closer coupling reduces stray magnetic fields, that can be more of an issue with a power transmission than with just a digital signal transfer in the old meters.

[iMo]

First time I can see TTL chips made in El Salvador.. Interesting..
PS: the transformer solution looks better to me - no LED brightness degradation over time..

[Kleinstein]

The transformer for digital signal transfer is relative bulky and needs extra circuit - however there are new transformer based couplers also in chip form.

[KK6IL]

The transformer for digital signal transfer is relative bulky and needs extra circuit - however there are new transformer based couplers also in chip form.

Some magnetic couplers, such as the NVE IL26x series, power up in an unknown state(BTDT). 5 pF isolation. The Silicon Labs Si866x does not require startup sequence, in to out C not specified.  Note those devices are for data transmission, not power, as are the transformers in the photo Vgkid posted.

John

[Kleinstein]

Those ring transformers shown by Vgkid may also be OK for low level power transmission, but in the 3456 and other old meters they were used to data transfer.

The 2 transformers shown in the picture are one side only - so there are 2 more similar transformers at the other side. So the 2 don't belong together, they are 2 separate data channels.

[iMo]

The transformer for digital signal transfer is relative bulky and needs extra circuit - however there are new transformer based couplers also in chip form.

There are about 4-5 TTL chips at the both sides usually - no problem to put all the stuff inside a smallest cpld or an fpga. Thus you would need the 2 small FT37-43 toroids (an example) and maybe 2 resistors to limit the current into the transfromer winding. With two cores per channel (one is at tx, one is at rx side) the coupling capacitance gets to a half and the max isolation voltage doubles..

[maxwell3e10]

I am wondering if anyone has suggestions for reducing small sharp spikes in a simple linear power supply that are associated with rectifier diode bridge. The spikes are synched to AC line and last 20-50 microsec. I am not sure how they propagate to the rest of the circuit, perhaps magnetic or capacitive coupling. They occur when the diodes turn on, so the exact phase depends also on the power supply load. Are there perhaps better diodes to use or putting some sort of filter to smooth out the turn-on transitions?

[Kleinstein]

There can be short current spikes from diode reverse recovery. This can be reduced by using fast revery diodes and limiting the voltage slew rate, e.g. with capacitance on the transformer output. Usually the diode turn on part is not an issue, and diode turn on pretty fast and not much difference (e.g. a 1N4001 is about as fast as a 1N4148 in that respect).
The other spikes come from the current pulse to the filter capacitor. Here some added resistance (could be just a fuse) or inductance can help to reduce the current pulsed and get some improvemen on the power factor. The fast rising current can otherwise be a problem and cause inductive coupling. The current puse may also be visible in the ripple voltage at the capacitors. 2 stage fitlering can reduce the shap pusle - it can also help to reduce the chance for a ground loop.
20-50 µs sounds about like the rising edge of the current pulse to charge the main filter capacitor.

[Echo88]

Heres an article that goes through the diode reverse recovery issue (if thats your issue in this case) in detail and shows how to solve it:

https://www.desmith.net/NMdS/Data/Linear%20Audio%20-%20Soft%20Recovery%20Diodes%20Lower%20Transformer%20Ringing%20by%2010-20X.pdf

[maxwell3e10]

Heres an article that goes through the diode reverse recovery issue (if thats your issue in this case) in detail and shows how to solve it:

https://www.desmith.net/NMdS/Data/Linear%20Audio%20-%20Soft%20Recovery%20Diodes%20Lower%20Transformer%20Ringing%20by%2010-20X.pdf

This looks promising! Adding resistor-capacitor network across the transformer already helped reducing the transient, just need to find optimal values. I will also get some SBYV27-200 diodes. Does anyone have other favorite diodes for general purpose linear rectifier?

[iMo]

In past while building ham radios I put 4n7-10n ceramics in parallel with each diode in the bridge rectifier..

[Kleinstein]

The difference between the different fast diodes is rather small - it looks like hard to measure that accurate (e.g. mains can vary over time). For low voltages ( keep in mind the diodes see up to 2 times the AC peak voltage) schottley diodes are attractive.  The choice depends on the current and availability. With only mains frequency there is no need to pick special or extra fast (e.g. SiC) didde - that is a point in the SMPS. The slow diodes have a Trr in the 2 µs range and even that is nearly fast enough (if mains is a clean sine) at mains frequency. Also series inductance (similar to PFC correction) can smoothen the current spike and thus make the reverse recovery less imortant.

[David Hess]

I am wondering if anyone has suggestions for reducing small sharp spikes in a simple linear power supply that are associated with rectifier diode bridge. The spikes are synched to AC line and last 20-50 microsec. I am not sure how they propagate to the rest of the circuit, perhaps magnetic or capacitive coupling. They occur when the diodes turn on, so the exact phase depends also on the power supply load. Are there perhaps better diodes to use or putting some sort of filter to smooth out the turn-on transitions?

I thought this was caused by reverse recovery "snap off", which some slow diodes display.  For 1N4004 series diodes, a couple hundred picofarads of capacitance directly across each diodes suppresses it.  Some designs include a ferrite bead in series with each diode.  Some standard recovery diodes have this problem and some do not.  Fast recovery diodes seem less likely to have this problem, or maybe they never do.

I first ran across this problem with a couple of simple fixed voltage power supplies that I had on my workbench for powering circuits during development.  Later I noticed that many linear power supplies in audio equipment include suppression capacitors across the rectifiers.  Tektronix included them in some but not all of their linear power supplies, implying that the problem was intermittent for them.

[Andreas]

The spikes are synched to AC line and last 20-50 microsec.

Hello,

It has already been said that radios and analog TVs used (up to 47nF) capacitors across each of the 4 diodes of the bridge rectifier.

some questions:
- how did you measure the spikes and with what amplitude?
  (on the AC-Side it would be very difficult so I assume that you measured across the smoothing capacitor on the DC-side)
- And what diode did you use for the bridge rectifier?

For my "ageing" cirquits I am using Schottky diodes (MBR20100) together with a dual winding transformer (2*12V).
Mainly to have not too much diode losses for the low drop regulator (15V).
But up to now I never had the Idea to measure for rectifier spikes.

with best regards

Andreas

[maxwell3e10]

Thanks for the suggestions. The spikes do in fact appear at the time of the turn-off of the diodes. The circuit in question is an old HP power supply, it has  a 50 nF capacitor across the transformer winding. The diodes used are 1N5059 which have maximum recovery time of 4 usec. I can probably do better with faster recovery diodes. The spikes are hard to see directly on rectified output, but they appear as ~0.5 mV spikes elsewhere in the circuit through some electromagnetic coupling.

[kleiner Rainer]

Everything you want to know about snubber design:

http://www.hagtech.com/pdf/snubber.pdf

Greetings,

Rainer

[David Hess]

Thanks for the suggestions. The spikes do in fact appear at the time of the turn-off of the diodes. The circuit in question is an old HP power supply, it has  a 50 nF capacitor across the transformer winding. The diodes used are 1N5059 which have maximum recovery time of 4 usec. I can probably do better with faster recovery diodes. The spikes are hard to see directly on rectified output, but they appear as ~0.5 mV spikes elsewhere in the circuit through some electromagnetic coupling.

When I had the same problem with my home built fixed output power supplies, I was working on audio circuits and the 120 Hz "buzz" showed up everywhere no matter what I did for shielding or filtering until I tracked the problem down to the standard recovery diodes and suppressed it at the source.

I added capacitance across each diode until the noise stopped, and then doubled the amount of capacitance.  The amount of capacitance needed seems to be proportional to the diode's capacitance, but it would not surprise me if it depends on the amount of stored charge.

Not all diodes of the same nominal type produced this problem.  It varied between batches of diodes from the same manufacturer, and different manufacturers.

[dietert1]

It also depends on the stray capacitance inductance of the transformer. I know that two chamber transformers with less perfect coupling between primary and secondary are much worse in this respect, independent of the type of diode.
A good/complete snubber has two capacitors and one resistor, as described in the document linked above by "kleiner Rainer". And it sits across each diode, not across the transformer secondary.

Regards, Dieter

[miro123]

Does somebody try gate driver dc-dc power supplies?
All major manufacture offer then. I have not seen any problems when I use then in intended for use application - high side SiC or GaN bridge? The only limitation is that you can use then only as single output dc-dc converter.
I don't have the equipment to measure 3pF parasitic capacitance.
https://www.murata.com/en-eu/products/power/isolated-gate-drive-power

[EC8010]

I see there's been lots of discussion about mains transformers and coupling. The key factor is C_PS (capacitance between primary and secondary). A standard foil electrostatic screen between primary and secondary on an EI transformer will reduce C_PS from 250pF on a 50VA transformer to around 5pF, perhaps 2pF. Adding capacitance to chassis from each end of the secondary creates a potential divider to common mode interference. Entirely covering the secondary with copper tape (without creating a shorted turn) reduces C_PS to about 0.1pF. It's fiddly work and requires a split bobbin transformer. Topaz Ultra-Isolator transformers fitted individual foil electrostatic screens around primary and secondary to achieve <10fF. They also have a safety screening plate between the two windings capable of sinking enough earth current to blow a fuse quickly. That's what it's there for; safety, not screening. Adding a safety screen to an EI transformer kit plus totally enclosing foil E/S screen on secondary, I achieved C_PS = 8fF +/-60% uncertainty. But what I really learned was that it's terribly easy to squander the improvement of a good transformer by sloppy external wiring. If you do something to prevent mains common mode interference getting in (low C_PS transformer, mains filter) you have to totally screen incoming mains. Folded tin plate screens work well over incoming mains wiring, with self-adhesive copper tape (gardener's slug tape) over all joints. RF crawls through the tiniest gap.

As has been mentioned, split bobbin transformers have high leakage inductance, needing an RC snubber across the secondary to damp their resonance when the diodes switch. Capacitors across diodes in the rectifier simply couple interference from the mains to your sensitive electronics. I know that's how it used to be done, but it was wrong. The Mark Johnson Quasimodo circuit works very well for determining the optimum snubber.

[alligatorblues]

If you want quiet, just get rid of the transformer. Take a metal box. Run some 120AC into it. Connect that to a 30W 500Ohm resistor with a thermostat in the circuit. Put the temp sensor in the metal box, set the thermostat for 60C, and put TEC chips in with the resistor. Put a cooler on the TEC chips, and you've got the cleanest power there is coming off the TEC chips: a thermoelectric generator. You just design the TEC pile to produce the voltage you need, and add the same configuration in parallel until you've got the current you need.

TEC power is cleaner than batteries.

[maxwell3e10]

Interesting idea, never heard of it being practically used though. Why do you say it would be cleaner than batteries? There will be some low-frequency noise due to temperature fluctuations and convection. Also one can still have some capacitive coupling to the power line coming in.

[Kleinstein]

Peltier elements / termolectric converters  have pretty poor efficiency. Even if they would run rather hot (e.g. 100 K temperature difference) the Carnot limit is only some 30% and the TE converts are something like 10% of this. So one can expect an efficientcy of less than some 3%.  A 30 W heater would thus hardly be good for some 1 W and one would need an hefty fan.

Chances are the a combination of IR LEDs and photovoltaic cells could be more efficient. At some 800-900 nm both the LEDs (laser diodes) and solar cells are relatively effcient (e.g. some 40% for the solar cell and lasers diodes).  There are systems that use power via optical fiber.

Another not much know option for isolation is an acoustic transformer: piezo elements connected with a glass rod.

For a voltage reference where size is not such an issue, there is also the option to power from batteries for the time of critical exleriments - normal standby over night could than use a conventional transformer. For those who run 24/7 the efficienty does matter a little.

[maxwell3e10]

I remember seeing a home-built HV power supply that was powered by a rotating plastic rod connected to a brushless motor. One can have a large distance between two parts and a clean AC signal at a few kHz to rectify for power. And the efficiency of motor/generator can be over 50%.