Analog ESR meter design

[Th3_uN1Qu3]

Re: Analog ESR meter design

If I were to design an ESR meter, this is how I would do it.

Constant current source pulses 1A thru cap, 100kHz @ 10% duty. Low-RdsON MOSFET (<5mohm) zeros cap on every cycle. Differential amplifier senses voltage across cap (through two less critical switches) and peak detector captures these.

This is where it gets tricky. For some reason, this works in simulation but does not work in real life. It creates a huge offset and makes the readings nonlinear... As i no longer have a scope i have no idea what actually happens, and this is the reason why i haven't been updating this thread.

[ben7]

Re: Analog ESR meter design

...As i no longer have a scope i have no idea what actually happens, and this is the reason why i haven't been updating this thread.

Aww, well I hope you can get one again sometime, I want one too.

China could add ESR in as a scale on any number of meters making ESR measurement essentially free. For some reason they don't.

I bet the crap-cap manufacturers don't want them to do that xD

[Th3_uN1Qu3]

Re: Analog ESR meter design

Aww, well I hope you can get one again sometime, I want one too.

I don't know.

My family situation is really complicated atm... i can't really afford anything anymore, i barely even eat these days. That's the very reason why i had to sell my scope in November - my mom got scammed and lost a significant amount of $$ ($2000, it may not seem like much to you but it means a lot in this country), so i was left to pay all the bills.

The situation is much worse now in that i hardly see her go to work anymore, and to add to that stuff has started breaking around the house (and i have to pay for it obviously), and i can't even find basic tools like the drill - i don't know where mom put it, and she's too lazy to look for it herself. I've looked everywhere i knew it could be, but no drill. Same for a bunch of other stuff. In this house, tools and money have a habit of vanishing... always did. I wanted to move out on my own this week but the weather sabotaged me. Until the snow clears and i pack my bags and move out, there won't be much i can contribute.

[Th3_uN1Qu3]

Re: Analog ESR meter design

I'm back at it. Found the drill in the meantime.

I fixed a couple groundloops in my messy perfboard design and now i get much lower offset when using 50Hz. Under 3mV with 1mV/mOhm resolution which is good enough for me. 50Hz isn't adequate for an ESR meter, but it sure works wonders at things like measuring resistance of wires and finding PCB ground loops.

However, when going past a couple kHz in frequency, the offset would still go crazy. At 67kHz i would get 265mV with the leads shorted. Since i added a separate wire going to the opamp (aka 3-wire measurement, it woulda been 4-wire but it's not a differential input so one of the wires is going to be ground anyway), i tried connecting the positive lead directly to the circuit ground and not to the probe going to ground. Surprise surprise - 70mV. It looks like i found one of the culprits... The wires i am using for the probes have very significant inductance at high frequencies. Gotta find wire with finer strands. The wires are from ATX PSUs btw, so now i've learned something new - some ripple filtering is done by the wires as well.

[ben7]

Re: Analog ESR meter design

I'm back at it. Found the drill in the meantime.

I fixed a couple groundloops in my messy perfboard design and now i get much lower offset when using 50Hz. Under 3mV with 1mV/mOhm resolution which is good enough for me. 50Hz isn't adequate for an ESR meter, but it sure works wonders at things like measuring resistance of wires and finding PCB ground loops.

However, when going past a couple kHz in frequency, the offset would still go crazy. At 67kHz i would get 265mV with the leads shorted. Since i added a separate wire going to the opamp (aka 3-wire measurement, it woulda been 4-wire but it's not a differential input so one of the wires is going to be ground anyway), i tried connecting the positive lead directly to the circuit ground and not to the probe going to ground. Surprise surprise - 70mV. It looks like i found one of the culprits... The wires i am using for the probes have very significant inductance at high frequencies. Gotta find wire with finer strands. The wires are from ATX PSUs btw, so now i've learned something new - some ripple filtering is done by the wires as well.

Wow, great observation!

-Ben

[Th3_uN1Qu3]

Re: Analog ESR meter design

Thanks. The original author recommended using a DMM with a relative function to eliminate the offset, but i'm pretty sure that's not the way to go - the offset is unpredictable to say the least. I'm thinking of using a differential amplifier to be able to use 4-wire measurement and completely take the probes out of the equation, which should result in zero offset.

Also there are nonlinearities at the lowest end of the measurement. A precision rectifier should take care of this. I think this works well enough to try and make a PCB now. In a later revision the 555 timer will be replaced with a more stable circuit that i just thought of, but i want to get the input and rectifier done first.

[Th3_uN1Qu3]

Re: Analog ESR meter design

I made some progress. Designed and built my circuit on a PCB, and it works... sort of. There is still offset produced by voltage drop on the wires even though i have a differential amplifier and separate wires for the current source and measurement input. I don't yet fully understand why. Not having a scope anymore really sucks. However, as opposed to the messy perfboard, on the PCB circuit the offset remains constant now and can be eliminated using the relative function of the multimeter. Oh and yes, it does work fine at 100kHz now, so it's a qualified ESR meter. I used a 110mOhm resistor for calibration, which i made from 10x 1.1Ohm 1% resistors, as this was the smallest value i found in 1% at the store.

The current source ended up dodgy, so in the current shape i need to divide the displayed value by 2, i'm too lazy to swap parts around right now. But given all this, the circuit works OK. It currently has 1mV/mOhm resolution (erm, well, 2mV), has a range of 1mOhm to 2 Ohm (that is if i fix the current source, now it's 1mOhm to 1 Ohm), and is accurate enough to spot caps that have even just started to go bad. I will add another range that goes from 10mOhm to 20 ohm, and if i find a way to get rid of the offset (or better wires) i'll also do 0.1mOhm to 0.2 Ohm.

I tested all my pulls (mostly junk brands, but a large portion of them are fine) and found over a dozen caps that were bad without bloating. I also found a bad Nichicon (likely simply of old age) i pulled out of a 19" Eizo CRT that had a blurry picture. Again, without bloating. If i had this meter when i decided to dismantle that, that monitor would still be working now.

[ben7]

Re: Analog ESR meter design

Hmm, I need to get a new breadboard but once I do, I can try to make one of these. I have some SMPS transformers, but I cant get the core apart without it breaking got any tips?

-Ben

[frako]

Re: Analog ESR meter design

I made some progress. Designed and built my circuit on a PCB, and it works... sort of. There is still offset produced by voltage drop on the wires even though i have a differential amplifier and separate wires for the current source and measurement input. I don't yet fully understand why. Not having a scope anymore really sucks. However, as opposed to the messy perfboard, on the PCB circuit the offset remains constant now and can be eliminated using the relative function of the multimeter. Oh and yes, it does work fine at 100kHz now, so it's a qualified ESR meter. I used a 110mOhm resistor for calibration, which i made from 10x 1.1Ohm 1% resistors, as this was the smallest value i found in 1% at the store.

You are using 4 wires for measurement? Are the two pairs properly shielded so no inductive coupling between current source and measurement pairs? You really need scope to debug that circuit...

[Th3_uN1Qu3]

Re: Analog ESR meter design

Yes i am using 4 wires for measurement. No, the wires aren't shielded, and the alligator clips i'm using really suck. I'll improve on that as much as i can and come with accurate results before i make the circuit public.

@ ben7: My circuit does not use a transformer.

[Th3_uN1Qu3]

Re: Analog ESR meter design

More progress. Good observation on the inductive coupling - i have separated the pairs of wires and the offset has reduced a lot. Gotta find some properly shielded wire. The other issue was the input impedance of my differential amp which was too low.

Currently the offset is under 3mV best case, but with it being so sensitive to the position of the wires the meter isn't really usable. I'll report back when i get shielded cable.

[cmj21973]

Re: Analog ESR meter design

Hmm, I need to get a new breadboard but once I do, I can try to make one of these. I have some SMPS transformers, but I cant get the core apart without it breaking got any tips?

-Ben

ferrite cores?
Transformers are dipped in lacquer.
Soak them in fingernail polish remover, acetone, or lacquer thinner overnight.

If the core is not broken into to many pieces, or any missing, it can be super glued back together and still work.

[Th3_uN1Qu3]

Re: Analog ESR meter design

Guys... This wiring stuff is nothing short of magic. I remembered i had some wifi coax from a scrapped laptop's antenna. I used that for my probes, connecting the shield only at the board end. The ground wire of the current source remained as it was as there wasn't enough coax left over.

By using this coax and grouping the four wires together as they leave the board, i got the display to show the magical zero with the probes shorted. No more offset. 0.0mV.

Right now the two wires going to the opamp are wired to the very tips of the alligator clips - those clips are absolutely awful. I'll have to get better ones. Oh, and i've yet to fix the current source, output still shows double what it should. I think i'll get on that right now.

Once i'm done with this i will release the circuit and PCB layout. For now i'll be sticking with 1mV/mOhm, should be plenty enough for most applications. Next challenge: 10mV/mOhm. Stay tuned.

[ben7]

Re: Analog ESR meter design

Guys... This wiring stuff is nothing short of magic. I remembered i had some wifi coax from a scrapped laptop's antenna. I used that for my probes, connecting the shield only at the board end. The ground wire of the current source remained as it was as there wasn't enough coax left over.

By using this coax and grouping the four wires together as they leave the board, i got the display to show the magical zero with the probes shorted. No more offset. 0.0mV.

Right now the two wires going to the opamp are wired to the very tips of the alligator clips - those clips are absolutely awful. I'll have to get better ones. Oh, and i've yet to fix the current source, output still shows double what it should. I think i'll get on that right now.

Once i'm done with this i will release the circuit and PCB layout. For now i'll be sticking with 1mV/mOhm, should be plenty enough for most applications. Next challenge: 10mV/mOhm. Stay tuned.

Awesome! glad to hear this is going well!

-Ben

[Th3_uN1Qu3]

Re: Analog ESR meter design

Attached the schematic in ExpressPCB format and as a picture for those who don't want to install the software. ExpressPCB is quite a crude package but i've gotten used to it and i can't be bothered to learn anything else right now. The PCB needs some updates, it'll probably get posted when i find a suitable case for the meter, i'll have to shop around next week.

C7 should be low ESR and can be of higher value. The probes absolutely must use shielded cable, with the shield connected to ground. If you want to make your own PCB before i post mine, the power side of the circuit (555 and current source) isn't too critical, but the opamp section benefits from a ground plane (connected to the ground of the opamp which is vcc/2, 4.5v in this case).

I decided to make the opamp section in SMD because i didn't find the MC33078 in a DIP package and i thought a SO-8 opamp with all thru-hole parts around it would look pretty lonely. I used 1206 and 0805 parts. I would have liked to have all passives in 1206 but that's how i found them.

If you want to use another opamp the requirements are 15MHz minimum bandwidth and low noise. The input offset doesn't matter since it's capacitor coupled.

[budm]

Re: Analog ESR meter design

Interesting, can the Constant current source be replaced by LM317? What is the current source output requirement? The OP amp output offset can be reduce if you can you can use precision voltage source for the 1/2Vcc or add adjustment pot to reduce the offset.
I see the output from the OSC, and two wires for the diff amp inputs, so how do attach the DUT to this setup. I may want to experiment with one and I want to be able to zero out the test leads itself also, I will more likely to use 1% resistors to improve CMR in the diff amp. Thanks for posting the diagram. I am thinking about building this and attach it to $5 DMM. I have SENCORE 75 this is sent out for Cal in yearly basis so I can use this to compare to what I will be making so I do not have to carry this big SENCORE around for small works.

[Th3_uN1Qu3]

Re: Analog ESR meter design

Interesting, can the Constant current source be replaced by LM317? What is the current source output requirement?

Yes of course it can be replaced by LM317. It's a textbook circuit i found over here: http://www.4qdtec.com/csm.html I chose that particular one because of the good thermal stability.

For 1mOhm/mV, the current required is somewhere in the 20mA range. Ideally it would be exactly 20mA but in practice this is going to be a bit higher due to the loss in the probes. Use a 100mOhm precision resistor for calibration.

The OP amp output offset can be reduce if you can you can use precision voltage source for the 1/2Vcc or add adjustment pot to reduce the offset.

The opamp output DC offset is essentially zero in this circuit as all inputs are AC coupled, and most opamps are trimmed for output offset, trouble happens usually when input offset is amplified. I could have gotten fancy and used another opamp to ensure a stable virtual ground, but in practice a pair of 1k 1% resistors and a polymer cap from virtual ground to power supply ground worked fine.

The offset i was talking about earlier, as frako pointed out, is created by inductive coupling between the current source wires and the opamp input wires, so the opamp is just doing its job, just that it's picking up what it shouldn't. In practice, this was the biggest challenge. Using wifi coax for the probes, with the shield connected to ground on the board end only, and grouping all four wires together as they leave the board eliminated this issue. Most important is the grouping of both opamp input wires as far as you can, as this makes all noise pickup common-mode and thus rejected by the diff input.

I used 1% resistors everywhere.

I see the output from the OSC, and two wires for the diff amp inputs, so how do attach the DUT to this setup. I may want to experiment with one and I want to be able to zero out the test leads itself also. Thanks for posting the diagram.

The output of the oscillator is between output and power supply ground. The inputs to the opamp join them as follows: IN1 to OUT, IN2 to GND.

This is intended for 4-wire measurement, so the inputs to the opamp should go as close to the tips of the probes as possible so the probes are compensated for. If you accidentally swap the inputs around you'll know it because you'll get a large offset and bad readings because of the power supply configuration - since this is a single supply circuit, the negative rail of the opamp is the ground of the power circuit.

[budm]

Re: Analog ESR meter design

Thanks for the explanations, I will try to build one using the LM317 for current source and using 1% resistors.
By the way, what do you get when you short all 4 test leads together, will you read the low resistance of the test lead it self, I am wondered if I can modify the circuits for checking for shorts in PCB since it can do really lo-Ohm measurement, I use my LC 75 for that function sometime.

[Th3_uN1Qu3]

Re: Analog ESR meter design

By the way, what do you get when you short all 4 test leads together, will you read the low resistance of the test lead it self

I get zero. That was what all the trouble was about... getting it to read zero with the test leads shorted was quite an undertaking, and i learned a lot about stray coupling between wires in the meantime. There are only two test leads (woulda been awkward to use four), but with two pairs of wires going to them.

For checking PCB shorts you would need to change the frequency to 50Hz and increase the coupling capacitor values so the 50Hz signal passes unaltered. Much less trouble that way. PCB traces are inductive too so if using a high frequency you could have the surprise that a short does not read as such unless the probes are really close to it. Another application for an AC milliohm meter working at 50Hz is as a "ground loop buster" in troubleshooting PCB layouts.

[budm]

Re: Analog ESR meter design

I will not be using the Osc section for the short finder, I will be using the Current source and some mod in the probe side for the shorts finder.