Topics

Looking for a better in circuit ESR meter


Reginald Beardsley
 

Please forgive the crosspost, but google and groups.io searches are not finding what I'm looking for. And I'm pretty sure the discussion was on one of these two lists.

There was a thread not long ago about in circuit ESR testers and someone linked to a very short (5-6 minute) YouTube video in which the designer demonstrated his design. In it he was marking bad caps with a marker. The tester in question was available from someone for around $239 IIRC. I think the designer's name began with "M". He also designed another very nice tester which is no longer available.

Distinguishing feature was it stepped through discharge and circuit resistance before evaluating the capacitor. I thought I'd saved a link, but apparently not. Maddeningly, I find messages from the thread I'm looking for but can't find the entire thread. and all I get from google is ads pretending to be reviews.

I'm trying to sort out a PSU problem with a LeCroy DDA-120 for which all I have is a schematic with no component values. Neither the Peak atlas ESR70 nor the BSIDE ESR02 Pro will work because of shunt resistors and diodes.

Thanks,
Reg


Reginald Beardsley
 

Found it. EDS-88A


Randy.AB9GO
 

I purchased one of those a few months ago and it is an excellent meter well
worth the money.

@AB9GO

This message sent to you from my mobile device via speech-to-text
technology.


Randy.AB9GO
 

PS. At one time I was doing the eBay try to find a good used one for less
money. Go ahead and buy the new one. There were several circuit updates
implemented later in the products lifetime. upon careful investigation
most of the ones listed on eBay do not have these updates. In addition the
older models are much harder on the battery. The only way to get decent
life is to run Eveready Energizer lithium cells and those do not come
cheap.

@AB9GO


This message sent to you from my mobile device via speech-to-text
technology.

On Fri, May 15, 2020, 9:05 AM Randy AB9GO <@AB9GO> wrote:

I purchased one of those a few months ago and it is an excellent meter
well worth the money.

@AB9GO

This message sent to you from my mobile device via speech-to-text
technology.


Renée
 

is this in reference to EDS-88A? pulled from Reg's comment?? or something else?
I love zero context
Renée

On 5/15/20 6:09 AM, Randy.AB9GO wrote:
PS. At one time I was doing the eBay try to find a good used one for less
money. Go ahead and buy the new one. There were several circuit updates
implemented later in the products lifetime. upon careful investigation
most of the ones listed on eBay do not have these updates. In addition the
older models are much harder on the battery. The only way to get decent
life is to run Eveready Energizer lithium cells and those do not come
cheap.

@AB9GO


This message sent to you from my mobile device via speech-to-text
technology.

On Fri, May 15, 2020, 9:05 AM Randy AB9GO <@AB9GO> wrote:

I purchased one of those a few months ago and it is an excellent meter
well worth the money.

@AB9GO

This message sent to you from my mobile device via speech-to-text
technology.


Reginald Beardsley
 

Does anyone have any experience with the B&K 881? It appears to provide the same testing as the EDS-88A for about $75 less. In fact looks like an outright copy with minor changes. Of course, that does not mean it is any good.

I'm dealing with a SMPS which has caps in parallel with a resistor and a diode. All crammed into a very small space. Both the Arduino based BSIDE ESR02 Pro and the Peak atlas ESR70 can't deal with the diode and resistor and produce similar <100 pF answers. From looking at the output on a scope I think both are turning the diode on.

Reg


Chuck Harris
 

You can always take the engineer's way out, and use your scope (OS)
and a function generator (FUNC).

......................................
FUNC o)=======@.......................
......|.......|.......................
......|......10K......................
......|.......|.......................
OS...o)=======@============@===========> To cap
......|....................|..........
......|................. DIODE........
......|..................CLAMP.......@=>
......|....................|.........|
.....///..................///.......///

Set the function generator to 100KHz, and 10Vpp, and set your scope
to external trigger on the function generator's trigger output.

The head to tail diode clamp will protect your circuitry from over
voltage by the function generator, and the scope will show a P-P
signal that represents the ESR. Since the current into the cap is
1ma, ESR is easily calculated from the P-P voltage measured.

Slow the function generator down to a point where Xc dominated the
impedance, and you can calculate the capacitance from the voltage.

It is ground referenced, so make sure the DUT isn't plugged into
anything that is grounded.

The idea is, of course, gratuitously stolen from others...

-Chuck Harris

Reginald Beardsley via groups.io wrote:

Does anyone have any experience with the B&K 881? It appears to provide the same testing as the EDS-88A for about $75 less. In fact looks like an outright copy with minor changes. Of course, that does not mean it is any good.

I'm dealing with a SMPS which has caps in parallel with a resistor and a diode. All crammed into a very small space. Both the Arduino based BSIDE ESR02 Pro and the Peak atlas ESR70 can't deal with the diode and resistor and produce similar <100 pF answers. From looking at the output on a scope I think both are turning the diode on.

Reg




Reginald Beardsley
 

Chuck,

I'm afraid I don't understand the diagram or the explanation. If there is a resistor in parallel with the cap, how much of the resistance is the resistor and how much is the ESR of the cap? You don't appear to address that. It appears to me you assume that there is no resistance in parallel with the cap.

If there is a diode and resistor in parallel with the cap, how are you going to keep the diode from affecting the result with a 10 V input?

It certainly seems to me that it is critical to not to forward bias the diode so that it conducts.

I've got a few dozen caps to check. That is what makes the EDS-88A so attractive. The caps are packed in so tight that lifting one leg is not possible and getting them back in would be a huge ordeal.

Please draw out a circuit consisting of a diode, resistor, inductor and capacitor in parallel and explain how you would separate the effects of the cap from the other 3 devices with a connection across the cap. That's the general in circuit test problem.

Have Fun!
Reg


Chuck Harris
 

Reg,

The "o)" symbol is a BNC connector. The ")" represents the grounded
shield on the BNC connector, the "o" represents the center pin.

One "o)" connector connects to the function generator, and the other
"o)" connector connects to the input on the oscilloscope.

The "@" signs are simply bends in the schematic wire, or nodes when the
wires cross or "T".

The 10K resistor, when fed with a 10Vpp 100KHz signal limits the current
to 1ma P-P.

The capacitor under test is connected where the two "=>" probes are on
the right side of the screen.

Ohms law is: V = I * R, and works equally well for AC, and impedance.

ESR is the series resistance a capacitor presents to the circuit.

Z = Xc + ESR

The idea behind I suppose all ESR meters is that at the frequency of the
measurement, the Xc of the capacitor is low enough that the ESR will
dominate the measurement.

As an example, let's assume that the capacitor we are testing is 1uf.

Xc = 1/(2pi * f * C) , Where f = 100KHz, and C = 1 E -6 Farads.

So, Xc = 1.59 ohms @ 100KHz

V = I * R, where I = 1ma, and R = 1.59 ohm.

So, as to the question of diode conduction, the expected voltage across
the capacitor is 1.59mv. Which is well below the conduction voltage of
the diode clamp.

A 10uf, or a 100uf would be even lower...

ESR + Xc values would have to be higher than 0.5V, or 500 ohms to
cause the diodes to conduct.

The 10K resistor makes your ESR readings come out in K ohms per volt,
so, 50 ohm ESR would be 0.05Vpp, and 5 ohm ESR would be 0.005Vpp...

If that is too far into the weeds, you can easily decrease the 10K
resistor to 1K, or even 100 ohms, to gain some more voltage in your
reading...

-Chuck Harris



Reginald Beardsley via groups.io wrote:

Chuck,

I'm afraid I don't understand the diagram or the explanation. If there is a resistor in parallel with the cap, how much of the resistance is the resistor and how much is the ESR of the cap? You don't appear to address that. It appears to me you assume that there is no resistance in parallel with the cap.

If there is a diode and resistor in parallel with the cap, how are you going to keep the diode from affecting the result with a 10 V input?

It certainly seems to me that it is critical to not to forward bias the diode so that it conducts.

I've got a few dozen caps to check. That is what makes the EDS-88A so attractive. The caps are packed in so tight that lifting one leg is not possible and getting them back in would be a huge ordeal.

Please draw out a circuit consisting of a diode, resistor, inductor and capacitor in parallel and explain how you would separate the effects of the cap from the other 3 devices with a connection across the cap. That's the general in circuit test problem.

Have Fun!
Reg




Reginald Beardsley
 

The DC resistance in parallel with the capacitor must be measured as it affects the AC measurement. If the diode conducts during either measurement, the results are not accurate.

This is what the B&K 881 and EDS-88A II do that. Nothing else seems to take into account. One of them uses a 25 mV signal for testing to avoid turning on a diode junction.

An inductor in parallel with the cap makes it rather messy. Fortunately, those don't appear in power supplies that I know of. They are usually series tuned to short the undesired frequencies to ground.

Have fun!
Reg


Chuck Harris
 

Lower the frequency, and you can easily measure the DC
resistance. If the waveform clips at the the clamp diode's
voltage, your DC parallel resistance is greater than 500 ohms.

Why should the diode conduct? It won't conduct until the
sum of the ESR and Xc is greater than 500 ohms.

If the diode conducts at 100KHz, your capacitor is bad.

-Chuck Harris

Reginald Beardsley via groups.io wrote:

The DC resistance in parallel with the capacitor must be measured as it affects the AC measurement. If the diode conducts during either measurement, the results are not accurate.

This is what the B&K 881 and EDS-88A II do that. Nothing else seems to take into account. One of them uses a 25 mV signal for testing to avoid turning on a diode junction.

An inductor in parallel with the cap makes it rather messy. Fortunately, those don't appear in power supplies that I know of. They are usually series tuned to short the undesired frequencies to ground.

Have fun!
Reg




David Berlind
 

Love this topic. Could someone redraw the diagram as an actual schematic that's posted as an image somewhere. I cannot make heads or tails of the diagram as is, even with the explanation. I think it has to do with how my computer is causing line-breaks and text-wrapping.


Chuck Harris
 

If my ascii schematic isn't working it is likely because you
have your browser set to substitute your favorite font for the
font chosen by the sender.

About the only time that is a good idea is if you want to see
the messages sent in a braile font.

In any case, the schematic is really too simple to waste much
time on.

-Chuck Harris

NODE #1 (F-Gen BNC center pin, 10K 1)
NODE #2 (10K 2, DUT probe 1, D1 1, D2 2, O-scope bnc center pin)
NODE #3 (F-Gen BNC gnd, O-scope BNC gnd, D1 2, D2 1, DUT probe 2)


David Berlind wrote:

Love this topic. Could someone redraw the diagram as an actual schematic that's posted as an image somewhere. I cannot make heads or tails of the diagram as is, even with the explanation. I think it has to do with how my computer is causing line-breaks and text-wrapping.




Reginald Beardsley
 

I'm using the groups.io website as my mailbox was getting swamped. I could make no sense of your ASCII art. Please get on the website and look at your own post. That's what I see.

I just ordered an EDS-88A II designed by Dave Miga from Merced Electronics. I'll see if it can successfully test the wad of caps that the BSIDE ESR02 Pro (aka Arduino LCR/transistor tester) and the Peak Atlas ESR70 cannot.

Both of those work for a lot of caps, but not for this tightly packed set of 12. They get reported as under 200 pF. Getting one back in after removal will be a huge pain. So I *really* want to be sure it's bad before removing one.

I'm trying to sort out a LeCroy DDA-120 which is misbehaving as if the front panel controls were intermittent. I suspect that I've got a noisy PS rail, but on an SMPS that's very hard to test because of ground access. I only have a few places I know of I can access the rails and no ground close enough to not have a huge problem with magnetic pickup.

Reg


 

Hi David,

Let me see if I can make a simple recount of what Chuck is explaining:

If you have an oscilloscope and a function generator able to output a sine wave of about 100 kHz, you have everything you need to make simple, quick ESR measurements in situ.

- Connect the ground and tip of the oscilloscope probe on both ends of the capacitor.
- Inject as a current the 100 kHz AC signal into the capacitor side that has the oscilloscope tip.
- Observe and estimate on the oscilloscope the voltage this AC current creates on the capacitor.

I set the function generator to 10 V, then connect this output to a 1 K resistor, and touch the capacitor with the other end of the resistor. This will create an AC current of 10 mA.
If the ERS of the cap is 1 ohm, the measured voltage will be 10 mV, something that can be recognized on the oscilloscope set to its high sensitivity of direct input.

It is not necessary to measure the value of ERS precisely, but only to get an idea whether the capacitor is good or not.
"Acceptable ESR" is also a function of the capacitance. In a large cap, a fraction of an ohm can be fine. In a small one, one ohm or larger can be acceptable.

Electrolytics are polarized. If you are a meticulous person, you will put the ground of the scope on the negative of the cap, and will see that the AC signal has a positive offset.
If you are a practical person, you realize that a small negative AC does not affect the response of the electrolytic cap. They are not like schottky diodes.

For this measurement, the less noise and less bandwidth your oscilloscope has, the better. And use a direct probe instead of a x10 probe. If you can trigger the scope on the AC signal (at the generator) like Chuck recommends, it is easier to evaluate the AC signal in the noise.
But the use of protective diodes may not necessary, since a few tens of millivolts with good ESR on the cap will not damage it, and several volts on a bad cap, won't make it worse. If the diodes are used to protect other components in the circuit, they won't interfere in the measurement because the signal on a good capacitor is so low.

Ernesto


Siggi
 

Here's a video by Alan Wolke explaining how this works:
https://youtu.be/115erzCCxgE. You want a square wave out of the
function generator, and the faster rise the better. With zero rise time,
the peak-to-peak voltage you measure is only a function of the capacitor's
ESR.

Chuck proposes to add a series resistor with the function generator, and a
back-to-back diode clamp across the capacitor under test to make sure
nothing gets fried.

On Sat, May 16, 2020 at 10:39 AM Ernesto <ebordon@...> wrote:

Hi David,

Let me see if I can make a simple recount of what Chuck is explaining:

If you have an oscilloscope and a function generator able to output a sine
wave of about 100 kHz, you have everything you need to make simple, quick
ESR measurements in situ.

- Connect the ground and tip of the oscilloscope probe on both ends of the
capacitor.
- Inject as a current the 100 kHz AC signal into the capacitor side that
has the oscilloscope tip.
- Observe and estimate on the oscilloscope the voltage this AC current
creates on the capacitor.

I set the function generator to 10 V, then connect this output to a 1 K
resistor, and touch the capacitor with the other end of the resistor. This
will create an AC current of 10 mA.
If the ERS of the cap is 1 ohm, the measured voltage will be 10 mV,
something that can be recognized on the oscilloscope set to its high
sensitivity of direct input.

It is not necessary to measure the value of ERS precisely, but only to
get an idea whether the capacitor is good or not.
"Acceptable ESR" is also a function of the capacitance. In a large cap, a
fraction of an ohm can be fine. In a small one, one ohm or larger can be
acceptable.

Electrolytics are polarized. If you are a meticulous person, you will put
the ground of the scope on the negative of the cap, and will see that the
AC signal has a positive offset.
If you are a practical person, you realize that a small negative AC does
not affect the response of the electrolytic cap. They are not like
schottky diodes.

For this measurement, the less noise and less bandwidth your oscilloscope
has, the better. And use a direct probe instead of a x10 probe. If you can
trigger the scope on the AC signal (at the generator) like Chuck
recommends, it is easier to evaluate the AC signal in the noise.
But the use of protective diodes may not necessary, since a few tens of
millivolts with good ESR on the cap will not damage it, and several volts
on a bad cap, won't make it worse. If the diodes are used to protect other
components in the circuit, they won't interfere in the measurement because
the signal on a good capacitor is so low.

Ernesto




Bert Haskins
 

On 5/16/2020 10:21 AM, Reginald Beardsley via groups.io wrote:
I'm using the groups.io website as my mailbox was getting swamped. I could make no sense of your ASCII art. Please get on the website and look at your own post. That's what I see.

I just ordered an EDS-88A II designed by Dave Miga from Merced Electronics. I'll see if it can successfully test the wad of caps that the BSIDE ESR02 Pro (aka Arduino LCR/transistor tester) and the Peak Atlas ESR70 cannot.

Both of those work for a lot of caps, but not for this tightly packed set of 12. They get reported as under 200 pF. Getting one back in after removal will be a huge pain. So I *really* want to be sure it's bad before removing one.

I'm trying to sort out a LeCroy DDA-120 which is misbehaving as if the front panel controls were intermittent. I suspect that I've got a noisy PS rail, but on an SMPS that's very hard to test because of ground access. I only have a few places I know of I can access the rails and no ground close enough to not have a huge problem with magnetic pickup.

Reg
1. !!!! Discharge!!!!! any/all caps.
2 connect scope and (F)unction (G)enerator ground to cap - / ground.

3. Connect function generator output to a 1K, 10k, or ? to protect FG output in case the CUT is shorted.

4 Set FG output to low.

5 Connect scope CH1 and the other end of the resister to the + side of the CUT.

6. Set FG to 1k, 10k, .. CPS and raise the output while watching the scope.

7 Keep the output of the FG ( as seen on the scope) to under .25 volts or so.


Needless to say, you can look at the FG output with CH2. and you will now have everything you need right on the scope screen.

I have used this method for decades mainly as as sanity check in conjunction with other methods.

Reply to this if you want to.

-- Bert


Miguel Work
 

I use similar method, but with FG and XY octopus and oscilloscope. I use 10R, 100R or 1K resistor, with a 100, 1khz and 10khz frequency. It works very well to test capacitors and transformers in circuit. Even I can put DC offset. +10 to -10, and frequency to 0 to supply little circuits and view current consumption.

Regards

Miguel

-----Mensaje original-----
De: TekScopes@groups.io [mailto:TekScopes@groups.io] En nombre de Bert Haskins
Enviado el: sábado, 16 de mayo de 2020 20:51
Para: TekScopes@groups.io
Asunto: Re: [TekScopes] Looking for a better in circuit ESR meter


On 5/16/2020 10:21 AM, Reginald Beardsley via groups.io wrote:
I'm using the groups.io website as my mailbox was getting swamped. I could make no sense of your ASCII art. Please get on the website and look at your own post. That's what I see.

I just ordered an EDS-88A II designed by Dave Miga from Merced Electronics. I'll see if it can successfully test the wad of caps that the BSIDE ESR02 Pro (aka Arduino LCR/transistor tester) and the Peak Atlas ESR70 cannot.

Both of those work for a lot of caps, but not for this tightly packed set of 12. They get reported as under 200 pF. Getting one back in after removal will be a huge pain. So I *really* want to be sure it's bad before removing one.

I'm trying to sort out a LeCroy DDA-120 which is misbehaving as if the front panel controls were intermittent. I suspect that I've got a noisy PS rail, but on an SMPS that's very hard to test because of ground access. I only have a few places I know of I can access the rails and no ground close enough to not have a huge problem with magnetic pickup.

Reg
1. !!!! Discharge!!!!! any/all caps.
2 connect scope and (F)unction (G)enerator ground to cap - / ground.

3. Connect function generator output to a 1K, 10k, or ? to protect FG output in case the CUT is shorted.

4 Set FG output to low.

5 Connect scope CH1 and the other end of the resister to the + side of the CUT.

6. Set FG to 1k, 10k, .. CPS and raise the output while watching the scope.

7 Keep the output of the FG ( as seen on the scope) to under .25 volts or so.


Needless to say, you can look at the FG output with CH2. and you will
now have everything you need right on the scope screen.

I have used this method for decades mainly as as sanity check in
conjunction with other methods.

Reply to this if you want to.

-- Bert


Richard Knoppow
 

With a lot of editing of the thread. I think I am missing something. I don't understand why ESR is so difficult to measure, it is measured by any impedance bridge although usually as Dissipation Factor. The difference is that ESR is independent of frequency while DF is a measure of the ratio of reactance to resistance at a given frequency. It is the reciprocal of Q. It is used because it is mathematically more convenient for capacitance but is exactly the same thing.
   Capacitors have both series and parallel resistance. Parallel resistance is often called Leakage. It is a measure of the direct current passed by a capacitor. A theoretically ideal capacitor should pass no DC but all practical ones do. Of course, the parallel resistance can also pass AC, lowering the effective reactance of the capacitor. So, one wants the lowest possible series resistance and the highest possible parallel resistance. Series resistance is strictly an AC measurement but parallel resistance can be either DC or AC. It may be difficult to distinguish series from parallel resistance for AC but for measuring the quality of a capacitor the DC measurement is sufficient for "leakage".
   An ideal capacitor has zero ESR and infinite parallel resistance. In fact, while ESR is a general AC measurement it  can be affected by losses at various frequencies so Dissipation Factor may be more appropriate to measuring the quality of a cap at specific frequencies.
    Of course, its well known that electrolytic caps have inherently high leakage because of the nature of the dielectric. They are usually used where the bulk value of the capacitance is more important than the quality, such as in power supply filters. The tolerance for the amount of DC passed, i.e., the leakage, depends on the application. It can also vary with applied voltage.
   It is possible to find capacitors which have reasonable ESR but excessive leakage. This is common in paper capacitors used as coupling caps where the capacitance will measure OK on a bridge but the DC allowed to pass will upset the bias on the following stage. Its also a problem for by-pass and decoupling caps in AVC circuits. Some capacitor checkers will measure both ESR or DF and leakage, typically impedance bridges will not measure leakage.

On 5/16/2020 11:51 AM, Bert Haskins wrote:

--
Richard Knoppow
dickburk@...
WB6KBL


Reginald Beardsley
 

The issue is doing it reliably in-circuit on an instrument for which there is very meager service information. All I've got for the PS is a schematic with no part values. Multiple caps in parallel are shown as a single cap. And access makes removing the caps for better testing a major chore. It would be very difficult to get them back in and with almost 20 caps wedged together down under the edge of major chassis parts, it's not at all clear it could be done without completely disassembling the supply. The construction is such that it would probably take 4 hours to take apart and another 4 hours to assemble.

The PS has 9 output voltage adjustments and 8 connections on the PS connector. I've not yet determined where I can access the -12V output. The service manual procedure is "replace module". Test points? We don't need no stinking test points!

The PS connector was poorly chosen and the +/- 5 V connections burn up. This unit had a factory repair under warranty. It consisted of soldering the PS to the main board with #12 AWG wire.

The symptoms make me suspect that I've got noise on a power rail, but the SMPS makes it very hard to probe for that.

The LeCroy DDA-120 functions, but the front panel response is intermittent. I've replaced the cable and the switch contact board and I've swapped the processor boards without significant change in behavior.

I'm Not Having Fun!
Reg