New low-cost ($170) 100MHz Differential scope probe from Micsig

[Hydron]

I did some quick measurements using a spectrum analyser + tracking generator to check the bandwidth myself. Not the best setup, but I did normalise it first and terminated the tracking generator with 50R (and measured across that).

Twisted (tw prefix) vs untwisted (untw prefix) input wires had a large effect, and both had a big peak before they started dropping off. I'd ignore the low frequency wobbles - I know they're pretty flat below 10MHz from other measurements.

I think it shows that beyond 50MHz it's a bit of a crapshoot with leads that long. Fluke made a probe (DP120) with resistive co-ax input leads - I'd be interested to try one of those to see if it's damped by the input leads, though it's only 20MHz too, which would make things much easier to calm down.

I am going to mod one of my probes (the design of the gain selection and diff-SE conversion sections leaves a LOT to be desired, though the input buffer looks OK) in the future, so might look at shortening the leads at the same time. Will probably do another post about that if i get time on the weekend.

[alm]

I think it shows that beyond 50MHz it's a bit of a crapshoot with leads that long. Fluke made a probe (DP120) with resistive co-ax input leads - I'd be interested to try one of those to see if it's damped by the input leads, though it's only 20MHz too, which would make things much easier to calm down.

I have that probe (with Agilent branding). I recall having to twist the leads for it to meet its bandwidth (or CMRR?) specification while testing it. I pretty much permanently have the leads twisted except for the last 30 cm or so.

I am going to mod one of my probes (the design of the gain selection and diff-SE conversion sections leaves a LOT to be desired, though the input buffer looks OK) in the future, so might look at shortening the leads at the same time. Will probably do another post about that if i get time on the weekend.

I like the form factor Tektronix used with a long BNC cable and short flying leads much better. I guess the long flying leads are more convenient for portable operation (put the amplifier in a pocket/bag, or tape it to the portable scope) or use on large equipment, but on the bench much shorter leads would suffice.

[boborich]

Looking forward to your mod.

I did the same with a noise source (SDG2122x) and MDO3000's RF input, get the same result. But without a preamp, the dynamic range is.....

Thanks for the input.

[bitseeker]

I already got the HVP-70 before the Micsig came out. I do like having 10X rather than 50X. However, it's great to see more products coming out in the $100-200 price range since my scope still has three more channels available.

[nctnico]

Meanwhile I have got a MicSig DP10013 as well and had a short play with it.
I tried to use it on one of my products with a MOSFET push-pull stage to generate high voltage square waves. I measured the gate drive of the high-side MOSFET so one end of the probe is at the output signal and one at the gate.




Not bad but I would have liked to have less clunky test clips so it is easier to clip them on SMT transistors or chips (SOIC housing).

I did a comparison with one of my Pintek DP25 probes using my network analyser (twisted wires and 50x setting).

The Pintek DP25 terminated with 50 Ohm:


The Pintek DP25  terminated with 1M Ohm:


The MicSig DP10013 terminated with 50 Ohm:


The MicSig DP10013 terminated with 1M Ohm:


It is clear the DP10013 doesn't have a 100MHz bandwidth (at least when driven from a 50 Ohm source). When terminated with 50 Ohm it has a -3dB bandwidth of approx. 85MHz. OTOH compared to the much more expensive Pintek DP25 the MicSig DP10013 shows a much flatter frequency response.

[Micsig_support]

Anyone tested the bandwidth of the probe?
There is a big peaking at 80Mhz starting from 60Mhz, and fall off at 9xMhz at -3db. I don't think this qualifies for a 100MHz probe.

I've seen the same effect with some other 100MHz differential probe.
The root cause for this behaviour are the probes test leads, they are simply too long for a 100MHz bandwith. Try moving them around or twisting them, this heavily influences the frequency response above 60MHz.
IMO a diff probe claiming 100MHz BW but using such long test leads is just nonsense. The probe circuit itself would do a nice 100MHz -3dB rolloff, but using the non-removable leads spoils it all. Same thing as with the ground clip of an ordinary x10 probe, they tend to resonate around 100MHz.

His answer can explain your test result.

[CustomEngineerer]

Anyone tested the bandwidth of the probe?
There is a big peaking at 80Mhz starting from 60Mhz, and fall off at 9xMhz at -3db. I don't think this qualifies for a 100MHz probe.

I've seen the same effect with some other 100MHz differential probe.
The root cause for this behaviour are the probes test leads, they are simply too long for a 100MHz bandwith. Try moving them around or twisting them, this heavily influences the frequency response above 60MHz.
IMO a diff probe claiming 100MHz BW but using such long test leads is just nonsense. The probe circuit itself would do a nice 100MHz -3dB rolloff, but using the non-removable leads spoils it all. Same thing as with the ground clip of an ordinary x10 probe, they tend to resonate around 100MHz.

His answer can explain your test result.

So you are admitting that your probe as supplied can't actually meet the 100MHz specs? Why is it called a 100MHz probe then?

[David Hess]

So you are admitting that your probe as supplied can't actually meet the 100MHz specs? Why is it called a 100MHz probe then?

Technically the probe does have a -3dB bandwidth of 100 MHz when the leads are twisted just right.  It just also has 10dB of peaking and the horrible transient response to match making it only useful for measurements up to 40 MHz which coincidentally is about the bandwidth limit for honest probes.

It is called a 100 MHz probe because Micsig is lying.

[David Hess]

Fluke made a probe (DP120) with resistive co-ax input leads - I'd be interested to try one of those to see if it's damped by the input leads, though it's only 20MHz too, which would make things much easier to calm down.

Old differential probes worked this way using special calibrated x10 or x100 passive probes for the input leads and the fastest example I know of is 150 MHz.  Preamble, who was bought by LeCroy, makes a stand alone 100 MHz differential probe amplifier like this.

The lower performance high voltage differential probes are typically used for off-line switching applications where more than 20 to 40 MHz is not required anyway.  Instead of bandwidth think in terms of rise time.  50 MHz is 7 nanoseconds and how many power switches are faster than that?

Further, the common mode rejection of a high performance probe like the Preamble would not be sufficient in these applications anyway to take advantage of their 100 to 150 MHz bandwidth.  If you want to make high common mode rejection measurements with that kind of bandwidth, then a different method is required.

[VA]

it only useful for measurements up to 40 MHz which coincidentally is about the bandwidth limit for honest probes.

Give me a link to other "Honest" differential probe even 20MHz bandwidth(That's enough for me) with a price tag up to $120?

[David Hess]

it only useful for measurements up to 40 MHz which coincidentally is about the bandwidth limit for honest probes.

Give me a link to other "Honest" differential probe even 20MHz bandwidth(That's enough for me) with a price tag up to $120?

I do not know of any at that price unless you look to the used market.

[Jay_Diddy_B]

Hi,

I did some similar measurements with the Tektronix P5200 and the SI-9000A. These results were shared in this thread:



https://www.eevblog.com/forum/projects/high-voltage-differential-probe-design-for-review/msg595063/#msg595063

You have to be careful with the source impedance when making these measurements. This model shows that:



Here are the results of the three configurations. In the last configuration I am using different damping resistors:



This is just from the input divider and the lead inductance. You can see how lead inductance, would mess with the frequency response.

I have attached the LTspice model.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

This picture really illustrates one of the limitations of these probes, and a trap for young players:



The negative pulse on the Gate waveform s is probably not real. It there because the differential probe does not have enough common mode rejection ratio, to reject the large dv/dt impressed by the switching.

It is always best to place both probes on the switch node, Source of the top MOSFET, There should be no signal.

CMRR is often more important than BW.



Regards,

Jay_Diddy_B

[nctnico]

Ahum... you are pointing to the output signal and not the one measured by the differential probe. I should have written that in the desciption but then again over 90V gate drive voltage is a bit much for any MOSFET.

[David Hess]

You have to be careful with the source impedance when making these measurements.

For high voltage differential probes, I think there is an assumption based on the commonly intended application of off-line switching power supplies that the source impedances will be very low.

The negative pulse on the Gate waveform s is probably not real. It there because the differential probe does not have enough common mode rejection ratio, to reject the large dv/dt impressed by the switching.

It might be real and it might not.  I have seen that sort of gate waveform produced from a combination of high impedance from the gate driver and the dI/dT combined with the source inductance and dV/dT combined with the reverse transfer capacitance.

Note: nctnico cleared up that this is not showing the gate waveform but I have still seen gate waveforms that looked like that.

It is always best to place both probes on the switch node, Source of the top MOSFET, There should be no signal.

CMRR is often more important than BW.

I agree; the common mode rejection ratio is often more important than the bandwidth.

This test should always be done before trusting the differential probe with the measurement to find out if the common mode rejection ratio is sufficient.

[Janne]

About CMRR.. is it reasonable to assume that any of the cheap diff probes (sub 300$) will not be able do 60dB up to something like 100kHz? I'm thinkin of getting one to measure gate drive signals, but without good CMRR I would waste my money on a cheap probe and then would need to get a better one after that.

[David Hess]

About CMRR.. is it reasonable to assume that any of the cheap diff probes (sub 300$) will not be able do 60dB up to something like 100kHz? I'm thinkin of getting one to measure gate drive signals, but without good CMRR I would waste my money on a cheap probe and then would need to get a better one after that.

High voltage differential probes are all about equally poor at best under ideal conditions.  They should have a specification and hopefully a graph showing the minimum common mode rejection ratio versus frequency but this assumes the specifications can be trusted.  For the best common mode rejection, the source must be low impedance which common gate drive measurements should be.

If this is important, then the test described above will reveal if it is a problem and if it is, then some other measurement method will need to be used.

The big obscure complaint I have heard is that the common mode rejection ratio of cheap differential probes drifts which is not unexpected given their construction and of course there is no service documentation about adjusting it.  Older expensive high voltage differential probes spend more effort in the design and construction of just their front end attenuators than is applied to an entire cheap differential probe.  (1) I have tested the common mode rejection of my 20+ year old 100 MHz 7A13s and they are still spot on.

(1) If the attenuators are built on an FR4 type of substrate, then they are suspect.  This is very difficult to get right because of material and supplier problems no matter how good the design is.

[Macman]

About CMRR.. is it reasonable to assume that any of the cheap diff probes (sub 300$) will not be able do 60dB up to something like 100kHz? I'm thinkin of getting one to measure gate drive signals, but without good CMRR I would waste my money on a cheap probe and then would need to get a better one after that.

The spec sheet for the probe states a CMRR of >60DB at 100khz but I would be good idea to wait until someone can do a test to confirm the real world figures.

I have purchased a set of 2 in the Batterfly deal because they seemed such good value. My only comments so far is that I would have preferred if the probes had been  X10 and X100, also the output seems a bit noisy buy I don't think that is uncommon for a differential probe.

What I would really interested to see done is a detailed comparison between the performance of Dave's HVP70 probe and one of these Micsig probes.

[Jay_Diddy_B]

Hi,

You have to remember that a 500x probe is -53dB, so if the probe has a CMRR of 53dB at frequency x MHz, then differential and common mode signal appear the same size.

Example:

Tektronix P5200

Differential Gain



Common Mode Gain




Regards,

Jay_Diddy_B

[nctnico]

I did a measurement on the DP10013 I have with both inputs connected to the 50 Ohm output and the output of the probe terminated with 1M Ohm. I get -84dB from 10Khz to approx 1MHz and -70dB at 10MHz in 50x mode (34dB). So it seems the CMMR is 50dB up to 1MHz and 36dB at 10MHz.

[BravoV]

Just curious can we get "better" signal integrity at probing, if it is designed with standard female banana input like at DMMs ? Instead of two dangling long and loose wires.

I was thinking something looks like this (from those Pintech probe) ...


And then coupled with isolated banana to an "isolated" female BNC and maybe with a cheap 10X HV scope's probe (no exposed metal too).

[nctnico]

Those isolated banana adapters look neat! I'm thinking about modifying my MicSig diff. probe with an isolated BNC plug so I can attach a 'normal' probe to it or just solder a piece of coax (with isolated BNC) to the DUT.

[2N3055]

It can't be be coax.. It must be twisted pair.. Both cables are "equal" but different polarity. Same impedance, capacity.....
Also, coupled interference has to to be induced equally in both cables so amplifier can cancel it out.. If you put one to be on coax shield , it will get more coupling than one protected inside... You could do twisted pair with outside common shield.. 

[BravoV]

Those isolated banana adapters look neat! I'm thinking about modifying my MicSig diff. probe with an isolated BNC plug so I can attach a 'normal' probe to it or just solder a piece of coax (with isolated BNC) to the DUT.

It a genuine old Fluke/Philips banana to bnc adapters, bought them while ago when they were cheap at the bay  (posted here -> Philips BNC/Banana converter, but nowdays Aliexpress also carries clones now, though not sure about quality compared to genuine Philips one (as its also convertible into non shrouded banana). 

Regarding isolated BNC plug, I bought this (attached below) few months ago at Aliexpress too, saw it at the 1st time, and instantly ordered it even I didn't need it, yeah, hoarder nerves kicked in. 

[BravoV]

It can't be be coax.. It must be twisted pair.. Both cables are "equal" but different polarity. Same impedance, capacity.....
Also, coupled interference has to to be induced equally in both cables so amplifier can cancel it out.. If you put one to be on coax shield , it will get more coupling than one protected inside... You could do twisted pair with outside common shield..

Ok, noted, thanks. 

Well, my intention or at least expecting that with banana plug instead of fixed wires, we can improve the probe and cabling, say like using shorter cables, twisted cheap wires soldered on the test points, or using dual coax cables with both shield say connected to ground earth and etc.

My point is, with banana plug instead of fixed wires, it opens up the flexibility and variations at probing the HV test points.