POWER DESIGNS PRECISION POWER SUPPLIES

[muvideo]

I tried to ask the seller if he wants to ship in europe, but ebay doesnt permit to contact him by
any means... maybe he answers no (probably, he lists the items for USA only) but I would have
contacted him directly

Fabio.

[robrenz]

Here are the CC results from my 2010.  200mA in current limit directly into the 8846A 400mA input with bnc coax and banana adapters. Set a 1V output.  Meter is in 6.5 digit mode and 100 NPLC.  AC current reading is 9.7uA rms so I think it would be 27.43uA peak to peak ripple assuming sinusoidal waveform.

This photo is the trend plot of the time to roughly thermal stabilize.  The total vertical range is 200uA and the horizontal is 7 hours 58 mins 40 secs. That is a long time to stabilize but I did turn on the air conditioning when I started. But 200uA drift over 8 hours with varying ambient is extremely good in my book. The m on the readings is mA.


This is the beginning of an 8 hour run on stat mode after just 9 readings. started right after the above 8hr stabilize.


These are the results after 8 hours and 16334 readings.  Temperature varied 2.4C with 25.2 max over the 8 hours.
Total min max span is 18.1uA with a SD of 2.658uA. Not bad for an ancient $60 dollar PS.


I will try the same thing on a 2005A next.

[robrenz]

I cleansed the rotary switch contacts and now it works really well. It's linearity is good but the unit is uncalibrated. But I'm happy. The only problem I have is the current limitation, its not so stable. If I look on the schematics, its probably temp drift in the shunt resistor or the pot, as they are "non ovenized".

MBY,  how unstable is the current? I am seeing about a 1.4mA variation on a 200mA setting on one of my 2005A's on initial testing. (not completely stabilized yet)

Regards, robrenz

[saturation]

Wow, I didn't know you could actually see the curve on the bench meter itself,  pics worth 1000s words.  Those are great results too, BTW.

This photo is the trend plot of the time to roughly thermal stabilize.  The total vertical range is 200uA and the horizontal is 7 hours 58 mins 40 secs. ...
I will try the same thing on a 2005A next.

[T4P]

Wow, I didn't know you could actually see the curve on the bench meter itself,  pics worth 1000s words.  Those are great results too, BTW.

This photo is the trend plot of the time to roughly thermal stabilize.  The total vertical range is 200uA and the horizontal is 7 hours 58 mins 40 secs. ...
I will try the same thing on a 2005A next.

Possibly a meter-not-a-scope-but-graph-display 

[MBY]

I cleansed the rotary switch contacts and now it works really well. It's linearity is good but the unit is uncalibrated. But I'm happy. The only problem I have is the current limitation, its not so stable. If I look on the schematics, its probably temp drift in the shunt resistor or the pot, as they are "non ovenized".

MBY,  how unstable is the current? I am seeing about a 1.4mA variation on a 200mA setting on one of my 2005A's on initial testing. (not completely stabilized yet)

Regards, robrenz

Embarrassedly (is that a word? Firefox says yes but I was never good in English back in school) I do not remember the values, but it was at least as bad as my "Voltcraft" LSP-1403 (a re-branded Manison cheap SMPS PSU), say 10 mA drift in 1 hour or so. My 2005 is in "cool shutdown" at the moment, but I fired the oven up just now. So, in a few hours I will check it again (the time is 01 AM here in Sweden, so the 2005 can warmup during the night (as can my DM3061 and Fluke 45), and I will check the current tomorrow).

A couple of days ago, I did run the 2005 for several hours to check the repetivity (again, is that a word? FF says no) of the knob settings (haven't touch them in several weeks), and it returned to 10.0006X VDC after half an hour and the value stayed there (see earlier posts for clarification).

[robrenz]

Wow, I didn't know you could actually see the curve on the bench meter itself,  pics worth 1000s words.  Those are great results too, BTW.

Trend plot is unbelievably usefull, I use it all the time. It is very sophisticated but super easy to use. Press the start button and the rest is fully automatic. Sample rate follows whatever the meter is set for.
Details here https://www.eevblog.com/forum/product-reviews-photos-and-discussion/looking-for-goodinexpensive-bench-power-supply/?action=dlattach;attach=24730
After having it I would not buy a bench meter without it.

Details of stats mode here https://www.eevblog.com/forum/product-reviews-photos-and-discussion/looking-for-goodinexpensive-bench-power-supply/?action=dlattach;attach=24729

One of the amazing things about this meter is the extreme internal resolution when measuring very small DCV deviations at 100NPLC.  When doing long term stability checks with a shorting bar the resolution in stat mode is 0.2nV  BTW the DCV stability over 152 hours with a 2.6 degC ambient variation is a total min max span of 1.08uV.

[robrenz]

So, in a few hours I will check it again (the time is 01 AM here in Sweden, so the 2005 can warmup during the night (as can my DM3061 and Fluke 45), and I will check the current tomorrow).

From what I have seen so far it appears it takes quite a while for the shunts in the meter and PS to thermally stabilize with a given current setting. The trend plot of the 2010 was 8 hours  before it was reasonably stable.  The drift in that time was 200uA. The drift after that was only 18uA.  So it appears that when you are doing current measurements in CC mode (that generate heat in the meter and DUT) it is easy to confuse instrument and DUT thermal stabilization with DUT long term drift.

[MBY]

So, in a few hours I will check it again (the time is 01 AM here in Sweden, so the 2005 can warmup during the night (as can my DM3061 and Fluke 45), and I will check the current tomorrow).

From what I have seen so far it appears it takes quite a while for the shunts in the meter and PS to thermally stabilize with a given current setting. The trend plot of the 2010 was 8 hours  before it was reasonably stable.  The drift in that time was 200uA. The drift after that was only 18uA.  So it appears that when you are doing current measurements in CC mode (that generate heat in the meter and DUT) it is easy to confuse instrument and DUT thermal stabilization with DUT long term drift.

Yes, there are lot of error sources. What I do now is to feed 200 mA through my DM3061 from another PSU to warm the measure shunt up. And my 2005 is "idling" on a setting of 10.000V with no load. My goal is to see the drift of the 2005 shunt/CC-circuit when "everything" _except_ the CC-circuits is loaded/warmed up.

[robrenz]

Excellent!  That measurement is more relevant to everyday use than the long term fully stabilized multi day stability.

[terabyte]

what are some practical reasons for a PS this stable?

[T4P]

what are some practical reasons for a PS this stable?

Some things that are extremely fussy about 200nV shift ? 

[SeanB]

Driving a couple dozen RTD devices in a plant, or driving an equal number of strain gauges, where you are feeding them all inot a converter, and need the readings to be consistent both with time and between sensors without having to calibrate individually.

[robrenz]

what are some practical reasons for a PS this stable?

Personally I find I gain a deeper understanding of a topic when I attempt to learn all the nuances involved in high levels of precision.  Being a machinist my real appreciation for high precision started when I bought my first electronic indicator with 1 micro inch resolution (38 years ago and I still use it frequently).  My eyes were opened when I could breath on a piece of metal and watch it grow or press on my lathe (which I always assumed to be infinitely rigid) with one finger and watch it deform.  Did I actually need that indicator when I bought it?  No, but the educational value and gut level understanding of how materials behave was invaluable. The same applies for me with electronics. I bought a 6.5 digit 24ppm accuracy bench meter. Do I actually need it? No but I am learning a lot by being able to see things happening that are not apparent with a regular meter.  As with all metrology, knowing what is involved in making accurate measurements at the bleeding edge of precision helps you improve the accuracy of your every day measurements.

These power supplies fascinate me because their old school analog design beats the pants off new high precision supplies. So all my posts here are me just tinkering around and learning by osmosis. Do I have an application that requires a PS with this level of precision?  No but I am learning a lot in the process.

[saturation]

A future of electronic devices is low power, low frequency, running off very small batteries if not alternative energy [ e.g. ultracaps] .  In order to design or troublehsoot such devices you need to be able to provide a clean stable source of 1-3VDC, capable also of bursts of high current, like a NiMH battery:  The PD design can output a stable 2A, 1.250 00 V very nicely.  You can do 1.25 V with most lab grade PSU but you'll have difficulty getting stability in the 10mV ranges, making accuracy difficult with very low voltage transducer outputs that resolve to 1mV, mA or even nV or nA range.

gas sensors: 02, c02, CO
thermocouples
strain gauges
output of biological sources like nerve fibers

emulation of low voltage batteries
kinetic energy sources of power/ charging [ e.g. body motion]
solar cells

http://www.kwjengineering.com/products/carbon-monoxide/sensors-for-carbon-monoxide/3et1co1500

http://www.ti.com/general/docs/litabsmultiplefilelist.tsp?&literatureNumber=slaa207

Before I found that such an ultrastable PSU existed, I used NiMH batteries.  Alas, in the design and test phase you do tend to use up more power than you expect until you tweak it down so as not to use up time charging them back up.

Sensors are expensive, expire, and need gas samples in order to get output.  With the PD supplies, I can emulate the output of many low voltage sensors that is usually done with VDC calibrator source.

what are some practical reasons for a PS this stable?

[saturation]

Yes, when you are working with very low power, the set up is very sensitive to ambient conditions, particularly if you are emulating a load with a resistor,  or a shunt.   For sensitive measurements, besides trying to control your climate conditions, I would not shut off the PSU or your DMM and leave them on for the entire period, i.e., days -months, you are anticipate doing testing, even if that just happens to be a few hours a day or a week.  In most metrology labs, such devices are left on perpetually!  FWIW I measured the quiescent draw of the 2020B in the 5W range.  That's $5/yr in energy cost.  My HP3456a draw about 25W each, about $25/yr.

So, in a few hours I will check it again (the time is 01 AM here in Sweden, so the 2005 can warmup during the night (as can my DM3061 and Fluke 45), and I will check the current tomorrow).

From what I have seen so far it appears it takes quite a while for the shunts in the meter and PS to thermally stabilize with a given current setting. The trend plot of the 2010 was 8 hours  before it was reasonably stable.  The drift in that time was 200uA. The drift after that was only 18uA.  So it appears that when you are doing current measurements in CC mode (that generate heat in the meter and DUT) it is easy to confuse instrument and DUT thermal stabilization with DUT long term drift.

Yes, there are lot of error sources. What I do now is to feed 200 mA through my DM3061 from another PSU to warm the measure shunt up. And my 2005 is "idling" on a setting of 10.000V with no load. My goal is to see the drift of the 2005 shunt/CC-circuit when "everything" _except_ the CC-circuits is loaded/warmed up.

[SeanB]

I leave my precision reference on for weeks at a time. Majority of the power is in the mains transformer ( about half) and the rest is in the 2 green LED's and the 7815 regulator that provides pre-regulation. The active regulator chip draws very little in operation. For me it is accurate enough to be used as a source of 10.000V for checking multimeters, at least on the 20V range. Other higher ranges with less accuracy, and I take the lower ranges on faith. Resistance I use some 0.1% resistors, and they are fine for the purpose.

[MBY]

Excellent!  That measurement is more relevant to everyday use than the long term fully stabilized multi day stability.

Exactly my thought!

Now I have measured and tested for hours, and I have some disturbing results. The CC on my 2005 is performing very poor, much worse than my cheap china PSUs.

Here what I did:
As I said, let my Fluke 45 and my DM3061 measure a current from another PSU to temp stabilize their current shunts. Meanwhile, my 2005 was powered on, dialed in (even max current), but with no load. Therefore, the theory is that every part in the measure setup is warmed to stability except the CC circuits in the 2005. After several hours, my cheap china PSU was delivering minimum of 150.4018 mA and maximum of 150.5309 mA:

Code: [Select]

Max Min Avg Smpl
LSP-1403 150.5309 150.4018 150.4949 17420
PD 2005 147.5291 145.7476 146.1012 100
PD 2005 147.5291 143.3662 144.3162 15970
PD 2005 145.1924 143.3182 144.0457 8820
(Every row with 2005 is a cleared of max-min, to see if things improve after warm up of the CC circuit)
The disturbing part is this: Max and min for the china PSU is drift, while max and min for the 2005 is short time oscillations. Even when I reset the statistics, in a few minutes, the max and min has reestablished.

But, when the CC knob is above the current depending on the load, everything is stable to sub µA and sub mV. So this does not seem to be a noise depending on the load, but on the CC circuits.

I also tried to measure open circuit V, small load V (without CC), with a scope, and even if the drift is very, very small, the noise is huge. I have a peak at 50 and 100 Hz (no surprises there) but the highest peek was at 1.59 kHz. The problem is that I have little or no control of noise-sources as I have my computer near, the scope is a USB one, etc, so the noise can come from everywhere. Anyhow, my 2005 is as good as everyone else's when as a precision voltage source, but my unit is useless as a precision current source. The oscillations does not seem to depend on temperature. Something else is the culprit. 

Sorry for my English, I hope you will understand my tests. Do anybody think that replacing ecaps is a good idea?

Edit: To be clear: All my gear has been on for several days now. The only thing that can be considered "cold" is the current shunt/CC circuits of the 2005. But the PSU has been ovenized and on for the whole time.

[robrenz]

From what I have seen so far on the 2005A the CC output is much more temperature sensitive than the CV output. My first step will be to replace the current sense resistor (R26 in the 2005A schem.)  that is in the CC control loop with a much lower resistive temp coefficient and higher wattage.  Just holding your hand on top of the PS for a minute will spike the current. The 12 mA droop in this picture is from a gentle blast of compressed air for just a few seconds directed at the top of the PS.

BTW your english is great.  It is infinitely better than my Swedish.

[saturation]

That's most execellent test, robrenz.  The current limit circuitry doesn't have the same precision passive components so it would be good to see what replacing it with better quality would do to avoid mA swings in current limiting.

[robrenz]

Wow, I didn't know you could actually see the curve on the bench meter itself,  pics worth 1000s words.  Those are great results too, BTW.

I started a separate thread on Trend plot to show more info and not get OT here. https://www.eevblog.com/forum/projects-designs-and-technical-stuff/trend-plot-on-fluke-884x-meters/msg114826/#msg114826

[MBY]

From what I have seen so far on the 2005A the CC output is much more temperature sensitive than the CV output. My first step will be to replace the current sense resistor (R26 in the 2005A schem.)  that is in the CC control loop with a much lower resistive temp coefficient and higher wattage.  Just holding your hand on top of the PS for a minute will spike the current. The 12 mA droop in this picture is from a gentle blast of compressed air for just a few seconds directed at the top of the PS.

BTW your english is great.  It is infinitely better than my Swedish.

Thank you! The trend plot is a real killerapp! Something I'm disappointed with regarding the DM3061 is the cheap stat functions. It shows max, min, average and number of samples only. Not even standard deviation, but every cheapish sci-calculator has one!

I have some good precision current shunts at home, maybe I will switch out the shunt in the 2005. But the thing that disturbs me is the "oscillations". The drift is no big deal (after a reset of the stats, the max and min values re-establishes, but the "window" does not seem to move. The oscillations is fast, maybe a couple of minutes. At first I thought it may depend on the oven thermostat, but my feeling is that the oscillations is faster than that. Any ideas?

Another thought: My 2005 may be a little bit underpowered. I'm using a 230->115, 50W transformer and the actual voltage is about 95 with the oven on, 92 VAC when oven on and full output load. But I still don't think the oven/thermostat is to blame for the oscillations since those are faster than the oven on-off cycle. But maybe the low voltage could be a problem anyway?

Speaking of oven on-off-cycle: It would be interesting to hear from others what they clock the cycle to, just for comparison? I haven't did that myself, but give me a few days and I will. (Amb temp + on/off duty and frequency after several hours)
The thermostat is presumably a bi-metal of some sort, as it gives a distinct sound. Have anybody played with the thought of replacing that with a PID control? I'm a little bit concerned with contact wear out.

[robrenz]

The resistance temperature coefficient is much more important than the precision of the resistor in this case. So make sure you know what those shunt tempco's are. I have looked and at low ohms and 3W it is hard to find anything with a tempco lower than 50 or 25ppm. I see those slow oscillations in CC also.  They are there in CV mode also but are very small.

I am pretty busy with work right now so I wont be participating much for the next week or two but I will be back.

Trend plots of oscillations when supply is set for 10uV.  That is 1.2886uV PP oscillation with spec. of 100uV PP  we are splitting hairs here but I do wonder what causes the slow oscillation.  The choppy plot may be aliasing from the low sample rate of the Trend plot.

[T4P]

Having a hard time ... deciding whether the fluke meter is a multimeter or a very slow scope 

[saturation]

Its like a scope's roll mode.  Several DMM that have trend capture have similar functions.





Its

Having a hard time ... deciding whether the fluke meter is a multimeter or a very slow scope