Improving Adjustable Dual Voltage Bench Power Supply

[xavier60]

Also remove the 270K resistor near pin 8 of the LM358.

I just removed it (it was the one marked 274) but now I no longer get any amp readings on the meter.

You could try putting it back but removing it should not have stopped it from working. The resistor is there to add some positive offset the LM358 which might cause a problem now. The voltages on pins 5, 6, 7 and 8 while there is some input might  tell me something.

[xavier60]

I think I know what's wrong. The meter needs to be zeroed. First see if you can get about 50mV on pin 7 of the LM358 using the Null Pot with the PSU unloaded. Ill look for a link that shows how to zero the meter,

[Frankentronics]

When I look at the datasheet for LM358 it says that pin 8 is the positive power supply for the IC. Is there a mistake?

[xavier60]

When I look at the datasheet for LM358 it says that pin 8 is the positive power supply for the IC. Is there a mistake?

Pin 8 is +3.3V supply to the LM358. The 270K resistor is connected from the +3.3V to pin 5, the Non-inverting input.

[xavier60]

 If you can't get 50mV on pin 7, put the 270K resistor back if you can still find it and then zero the meter again.
I didn't think all this through properly.

[Frankentronics]

OK, as per your instructions, I did all the steps and I reinstalled the 270k resistor. Then I did some calibration adjustments on the breadboard.

To be honest, I am actually very happy about how the circuit is performing. The readings are matching almost 100% and I guess I could continue tweaking until it's even more accurate, but I decided I might as well do it all on the final board. So, as we speak I am prepping the board then I'll populate it with all the components.

I'll post photos of the progress and I'll post an updated schematic. I'm actually doing two things at the same time so it takes a while.

But I have a technical question about the tapping of the 100k resistor into the #5 pin of the LM358. Is this basically bypassing the internal shut and also converting the amp meter into a volt meter?

Thanks...

[xavier60]

I'm still curious to see if D2 is still needed.
It could be seen that adding of the 100K converts it into a voltage meter and it could be used for one.
In this situation, the PNP transistor outputs purely current and that's what the meter is measuring.
The 100K resistor just protects the input. It could be bridged out or changed to 1M without affecting the calibration.
The input signal is combined with the internal CS signal and will still function normally.

[xavier60]

On your PCB make provision for a small electrolytic capacitor in place of R9. Because you will be powering audio circuits which draw rapidly fluctuation current, some extra smoothing might be needed.

[Frankentronics]

I didn't catch your post in time to include a place for the electrolytic cap. But it's actually very easy to add one, on this type of PCB. I already figured out how and where to place one.

Should it be 10uF?

Here are some pictures of how I made the PCB, using a dentist's drill bit, just like I made the main board. There are a lot of components used here that were pulled out of discarded electronics, both trim pots, the 100m Ohm resistor, the transistor, and the 100n cap. The rest are new components.

I'm still working on the revised schematic, but I'll have to finish it tomorrow.

I did test the circuit on the breadboard without the diode and basically the circuit worked, but I remember getting slightly different voltage measurements across R9. I'd be happy to run a test tomorrow and post the data, as I did on a previous test. I can easily take the diode out of circuit by just putting a jumper with alligator clips across the diode, right?

[xavier60]

 Your PCBs turn out surprisingly ok considering.
D2 can be jumpered out  for the test. Try a TL071 if you get the opportunity.
A 1uF or a bit less between the PNP transistor's collector and ground should be more than enough smoothing without affecting the reading settling time by too much. A monolithic cap might be better, possibly less leakage.
What resistance did R8 need to be?  Im also curious if you took note of how much the amperage reading could be changed  with the Null Pot alone.

[Frankentronics]

Try a TL071 if you get the opportunity.

If I have one I think I will build another breadboard circuit and see how that one performs.

What resistance did R8 need to be? 

It's a bit odd because on the breadboard I just put a 10k trim pot in series with the 10k resistor and I was able to adjust so that the reading of 500mA matched. I assumed that was going to be the case on the PCB but I still made 3 spots for resistors, just as you've suggested. Then, to my surprise, I was not able to match the reading at 500mA, even when I dialed the trim pot all the way down. So I added another 10k resistor in parallel with the existing one, so that I could dial up the trim pot. That worked out fine.

I can't imagine that this was due to subtle differences in the resistance of the PCB traces compared to the internal resistance of the breadboard. There must be some other explanation.

Im also curious if you took note of how much the amperage reading could be changed  with the Null Pot alone.

Why don't I just build another circuit on the breadboard and I can post all the readings that you are interested in. This worked out really well for me and for sure I could not have done it without your help. I'd be more than happy to post any measurements that you are interested in or test any variations of the circuit.

There is no solution for this that I could find on the internet and I'm pretty sure no one ever figured out any solutions for using these panel meters on a dual power supply. I am really impressed with your skills and forever grateful for your help.

Thank you very much.

[xavier60]

Eventually when you understand how the circuit functions, you will be able to calculate the effect of component values such as Rcs and R8.
The transistor current can be calculated from the voltage across the 100K.
I got the idea from the way constant current sink circuits work, often discussed in this forum.
 Although it has a different purpose, the operating principles are the same.
https://electronics.stackexchange.com/questions/129609/how-does-this-constant-current-sink-actually-work

Ill stand by.

[xavier60]

I did some tests on a TI branded LM741 I found here. The Null Pot gave a rather symmetrical input offset range of +/- 10mV.
This would result in +/- 100mA change in current reading.
The LM741 isn't happy to have its input pins at + supply voltage,  200mV below is enough for it to work properly. 
Some rough calculations indicate a value for R8 of about 6.8K.

[Frankentronics]

I plan to do more testing today and post the results. It should be interesting.

Perhaps once this is completely finalized and after I've used the power supply for a while I might build another one and put all the circuits one a single PCB. It can't hurt to have two power supplies.

There is one question from an earlier post that I did not resolve. Here is how I connected the volt meters.



My question was about that connection that appears in the lower right corner of my schematic. As you can see there is an isolated power supply that I am using to power up the panel meter that is used to measure the negative side of the main power supply. In order for that meter to be powered up the thin black wire had to be connected to the ground of the secondary power supply. However, in order for that same meter to measure negative voltage of the main power supply that same wire had to be also connected to the negative output rail. It works and I don't think there's a problem, but I'm not really sure, so I ask.

Are there any issues to doing it that way?

Thanks...

[xavier60]

I would use the thin red and black wires just to power the meter from the isolated supply.
Connect the thick Black to the main negative output from the PCB.
And connect the thick Red to the negative output terminal on the front panel.

[Frankentronics]

I would use the thin red and black wires just to power the meter from the isolated supply.
Connect the thick Black to the main negative output from the PCB.
And connect the thick Red to the negative output terminal on the front panel.

Sure, but since the thick black and thin black are internally connected, the ground from the secondary supply will still be connected to the negative output of the main supply, correct?

Is there a technical issue with that?

Also, I did some preliminary tests, jumping the diode, using the PCB I just built. Let me explain what the photo shows, first.



So, I basically connected the second panel meter to its independent power supply and then I connected that secondary meter with my DMM in series with an 8 Ohm resistor. The modified panel meter is connected through the PCB and there is an alligator jumper connected across the diode, making it our of circuit.

The modified meter does not track the readings. If I adjust the gain trim pot to match a higher reading (at 500mA) then when I dial down the output, the meter will not track. It will zero out around 330mA. Also, no amount of turning the null trim pot will fix that.

I will do more tests and also build a new circuit, but I have to fist find all needed components.

[xavier60]

Although the thin and the thick black wires are connected internally, in some situation it's best not to join them externally. It can alter readings.
Tracking can be tricky. Try this procedure. disconnect the input to the 100K resistor. Use the instructions in the video to zero the meter's current range. I had one that wouldn't keep its new zero calibration so test it by seeing how much voltage has to be applied to the 100K resistor to get  readings of 0.02, 0.05 and 0.50, then let me know.

If you cant do this test easily, just connect it back up again and see what effect the Null Pot has on the reading. Dont load the PSU.

[Frankentronics]

Try this procedure. disconnect the input to the 100K resistor. Use the instructions in the video to zero the meter's current range. I had one that wouldn't keep its new zero calibration so test it by seeing how much voltage has to be applied to the 100K resistor to get  readings of 0.02, 0.05 and 0.50, then let me know.

Will do.

If you cant do this test easily, just connect it back up again and see what effect the Null Pot has on the reading. Dont load the PSU.

You mean, when I put R9 back into the circuit, correct?

Also, I found a 1uF film cap. Should I just place that cap in the circuit even when R9 is connected?

Thanks...

[xavier60]

Try this procedure. disconnect the input to the 100K resistor. Use the instructions in the video to zero the meter's current range. I had one that wouldn't keep its new zero calibration so test it by seeing how much voltage has to be applied to the 100K resistor to get  readings of 0.02, 0.05 and 0.50, then let me know.

Will do.

If you cant do this test easily, just connect it back up again and see what effect the Null Pot has on the reading. Dont load the PSU.

You mean, when I put R9 back into the circuit, correct?

Also, I found a 1uF film cap. Should I just place that cap in the circuit even when R9 is connected?

Thanks...

Dont bother with the capacitor for now. R9 was used only for initial testing of the circuit and was supposed to be left out when the modified meter got connected. The Collector of the PNP transistor should only connect to the 100K resistor in the meter.
It's important that when the meter's current range  is zeroed, there is no input to the 100K resistor.

[xavier60]

I might have misunderstood something, it was the middle of the night when I read your posts.
The poor tracking was being caused by the bridging out of D2 test. So there is nothing really wrong except for R9 still being in circuit?

[Frankentronics]

I might have misunderstood something, it was the middle of the night when I read your posts.
The poor tracking was being caused by the bridging out of D2 test. So there is nothing really wrong except for R9 still being in circuit?

I think both of us misunderstood each other. And it was in fact me that first misunderstood your post when you said this...

...test it by seeing how much voltage has to be applied to the 100K resistor to get  readings of 0.02, 0.05 and 0.50, then let me know.

If you cant do this test easily, just connect it back up again and see what effect the Null Pot has on the reading. Dont load the PSU.

For whatever reason I thought you wanted me to put R9 back in circuit and do the calibrations with the diode, then do the test. That's why I mentioned R9 and then you thought that I forgot to remove it. But I did in fact remove it as you can see if you look back at the photo I posted in my reply #46. Yo can see how the R9 resistor was cut off, right under the transistor.



In any event, I reread you posts and now I understand what you meant. So, to start with I made a log of how many volts I need to apply to a 8 Ohm load to get certain mA readings, done with the diode in circuit and both trim pots properly calibrated. Here is the data.

0.00 A @ 0.0 V
0.01 A @ 0.0 V (I guess the meter doesn't register less than 40mV change)
0.02 A @ 0.1 V
0.05 A @ 0.3 V
0.10 A @ 0.7 V
0.20 A @ 1.6 V
0.30 A @ 2.4 V
0.40 A @ 3.3 V
0.50 A @ 4.1 V

All of the above Amp readings matched perfectly what I was simultaneously reading off of my DMM.

Next test I jumped the diode D2 and logged how many volts I need to apply to the 8 Ohm load to make 0.02, 0.05 and 0.50 Amps show up on the meter. I did not change the calibration of the trim pots. I also logged the actual amp readings off of my DMM.

0.02 A @ 1.1 V (reading from DMM 0.14 A)
0.05 A @ 1.2 V (reading from DMM 0.15 A)
0.50 A @ 4.5 V (reading from DMM 0.56 A)

Then for the next test I disconnected the jumper across the diode and re-checked that if I was going to get the exact same readings as I did on my first test. At this point I still did not touch the trim pots. The readings were dead on, as on the first test.

Then for the next test I jumped the diode again, but this time I calibrated the gain trim pot R8 to display 0.50 A when the DMM was also displaying 0.50 A. Of course, since this was the correct current and I have not changed the resistor it needed 4.1 Volts. Here is the data.

0.02 A @ 1.4 V (reading from DMM 0.18 A)
0.05 A @ 1.4 V (reading from DMM 0.19 A)
0.50 A @ 4.5 V (reading from DMM 0.56 A)

You will notice that both 0.02 A and 0.05A needed 1.4 Volts. Well, I guess the change was just a few milivolts and the meter didn't register. In fact, I had to breadboard a fine adjustment 1K pot in series with my 4.7K pot, so that I could hit the fine adjustments between 0.02 and 0.05. It was simply impossible to dial in that level of accuracy with the coarse 4.7 K pot. However, in the final product I will be using a multi turn pot on the front panel.

I hope those measurements are helpful.

I still didn't have time to breadboard a new circuit cause I got pulled away from my workshop. I'll do it tomorrow. I'm also curious to see how the 071 Op Amp will behave.

Thanks...

[xavier60]

All that looks reasonable and expected. There might be a typo in the first table, 2nd line. I notice with all the meters that I have received, they are incapable of displaying 0.01A, they jump directly from 0.00 to 0.02. Whatever the programmer has done also causes an inherent zero offset error.
If an unmodified meter is properly zeroed then calibrated at some low current, at higher currents it will read higher than the actual current.
You wont have this problem as you are correcting the zeroing error with the Null Pot.

 You might get some drift caused by the LM358. Try gently flexing the meter's PCB to see what happens to a current reading.
With your present experience level, it is too risky to remove the LM358 for now.

[Frankentronics]

There might be a typo in the first table, 2nd line. I notice with all the meters that I have received, they are incapable of displaying 0.01A, they jump directly from 0.00 to 0.02.

I was actually able to get 0.01A on the display. Here I am including a photo that shows the 0.01 display. I was only able to hit this number because I asses a fine adjustment pot in series with the output voltage adjustment pot (seen in second picture). But even so, perhaps the programmers updated the firmware for the unit. It's possible they were receiving complaints and decided to debug the code.

With your present experience level, it is too risky to remove the LM358 for now.

I actually feel quite comfortable removing that chip. I have an SMD heat gun and I've been practicing on discarded electronics for years.

What chip would be better suited for this? And should it be replaced in both meters?


Also, I finished the updated schematic.



Now I'm working on another schematic showing how the meters are hooked up.

And I'm also still working on breadboarding a second circuit for testing.

Thanks...

[xavier60]

Where did your meters come from? I would like to try some. I did notice that some of the resistor values are different to mine.

[Frankentronics]

Where did your meters come from? I would like to try some. I did notice that some of the resistor values are different to mine.

I bought this one...

https://www.banggood.com/Mini-Digital-Voltmeter-Ammeter-DC-100V-10A-Panel-Amp-Volt-Voltage-Current-Meter-Tester-0_56-Blue-Red-Dual-LED-Display-p-1422461.html?rmmds=search&cur_warehouse=CN

...however, one never knows if they send you the exact same model.