Improving Adjustable Dual Voltage Bench Power Supply

[xavier60]

That makes sense. I assumed that there would be a problem in wiring the positive meter to read current properly. Maybe I'm over thinking it.

If I understand correctly, there is in fact a problem wiring up the positive for current. I found another thread on this forum...

https://www.eevblog.com/forum/beginners/digital-panel-meters-measuring-negative-voltages/

...with the following image.



Clearly that would not be what I need because I would need the 0V output to be a single terminal.


I think I am now better understanding your schematic for the current to voltage converter. Although I still don't fully understand how to wire it all up. I did some experiments, only for the voltages, but they all failed. It seemed like this was going to be a straightforward thing but once it came down to figuring out how to make the connections it's not as easy as I had assumed. And I'm really not finding any solutions through Google.

There should be no problem with measuring voltage and current for the negative output. The only complication is that the 5V regulator needs to be supplied with about +7V.
The meter for the positive output should be ok just measuring voltage for now, powered from U1.

[Frankentronics]

There should be no problem with measuring voltage and current for the negative output. The only complication is that the 5V regulator needs to be supplied with about +7V.
The meter for the positive output should be ok just measuring voltage for now, powered from U1.

OK, this time I understand everything that you just explained.

I guess the question that remains is how to resolve the current for the positive output. Is there a solution?

Thanks...

[xavier60]

There should be no problem with measuring voltage and current for the negative output. The only complication is that the 5V regulator needs to be supplied with about +7V.
The meter for the positive output should be ok just measuring voltage for now, powered from U1.

OK, this time I understand everything that you just explained.

I guess the question that remains is how to resolve the current for the positive output. Is there a solution?

Thanks...

Yes, read from Reply #16.
The current to voltage converter will likely need some tweaking depending on what type op-amps you have?
Then the meter will need a simple mod. I had to do a lot of hacking to make one work properly with my bench supply.
https://www.eevblog.com/forum/projects/linear-lab-power-supply/msg2409717/#msg2409717

[Frankentronics]

I did some tests on the breadboard. The diagram that was posted in the other thread, drawn on the white board, did not work, because of how that 7805 is wired, by being tapped behind the 337. Even when I tapped in front of the 337 it did not work.

I decided to test the voltmeters first and worry about the amp meters later.

So, I powered up the positive voltmeter straight from the 7805 that was already on the board that I made (the one that's powering the 741). Then I wired a separate 7805 to provide separate powering for the negative voltmeter. I utilized the salvaged rectifier board, which in fact has two rectifiers. However, there was an internal connection on that salvaged board, which connected the ground of the rectifier 1 with the ground of the rectifier 2. I broke off that internal connection.

The voltmeters worked but I'm not sure if they worked as they should. The negative voltmeter seems to have a reading that is slightly off (only by 0.1 volt). But the real question for me is that I had to connect the ground from rectifier 2 with the -Ve output from my power supply board. And I just don't know if I'm supposed to do that. In fact, that's the reason why I severed that internal ground connection, on the salvaged factory rectifier board.

I made a schematic to explain what I mean.

[Frankentronics]

Here is one solution. It's a current to voltage converter that level shifts the signal voltage across the high-side CS resistor to the low-side.
The output can be fed into the I_sense A/D pin of the meter's micro-controller.
The performance is dependent on the op-amp's characteristics. Most important is that it will work properly with it's inputs at + supply voltage.
I think the LM741 is a suitable starting point. TL071 also.

I studied your circuit a bit and tried to gather components for testing.

First I made a schematic that shows how I think your current to voltage converter circuit should be connected into my circuit. Is my schematic correct?



Also, I can't find any BC558 transistors. But I did find a few PNP transistors in my salvage bin. Can you tell me if any of those are suitable replacements for the BC558?

I found the following ones: BD140, 2N3906, S8550, 881Y, C327, B331, A 1296 (I do have more but I have to go through them).

You mentioned that the amp meter needs a minor mod. Would that be removing the internal shunt?

The way I understand the am meter should be connected is to connect the thick red wire to the OUT of your circuit and the thick black wire to the 0V. Is that correct?

I have some .22 Ohm resistors. Would two of those in parallel be a good enough replacement for that 100m Ohm resistor?

I am looking for a 12V zener. Would any 12V zener work or do I need to know any other specs for the zener?

Thank you...

[xavier60]

The load resistor R9 converts the current signal coming out of the Collector of the PNP transistor into a voltage just for testing for now.
When the circuit is eventually connected to the meter, the load resistance will be in the meter.
This project shouldn't be too difficult to get working in a fashion. The hard work will be in refining it to get acceptable accuracy.
Does your meter look like this?
https://www.sunrom.com/media/content/2260/vc288-500x500.jpg
There are a few ways of getting the current signal into the meter. One safe way involving no mods is to make a connection to pin 5 of the LM358 via a 100K resistor.
The PCB is tricky to get out of the housing without breaking off a retainer tab. At the shunt end, I wedge small screw driver between the edge of the PCB and the tab and gently pry open the gap until the protrusion at the end of the PCB is just clear of the slot. Another screw driver is then used to lever up the PCB by the shunt. I then file down the protrusion to make the PCB easier to remove next time.
Get the circuit working well enough first.
Only the 3 wires are needed to be used on the meter plus the current signal wire.

[Frankentronics]

Thank you, again, for the information.

Correct, my meter is the same model. I went ahead and installed that 100k resistor on pin 5 of the LM358. So the meter is prepped.



I first glued the resistor in place with rubber cement and when it was dry I soldered the connection.

I also found all the necessary components for the circuit. I researched transistors a bit and found out that the 2N3906 should be a suitable replacement for the BC558. I also found a 1W zener diode 1N4742A. The only component I do not have is the 100m Ohm current sensing resistor. But I do have two .22 Ohm resistors that I connected in parallel. So the circuit is now built on the breadboard.



If I understand correctly I'll have to cut the trace on my circuit board, in front of the LM317 and tap the board there to run this circuit. I am not 100% clear on how to proceed with the testing, though.

Thanks...

[xavier60]

Ideally the voltage across R9 should be 0V and then increase linearly with increasing voltage drop across Rcs.
This depends all on the op-amp's specs. The output needs to be able to swing high enough to completely turn off the transistor. If it cant, it will need a 10K pull up resistor from its output pin to + rail.
Next is that the op-amp can work normally with its input pins at + rail voltage. I know that the LM741 can have its inputs very close to + rail but I don't exactly know how close. If a problem is suspected,a temporary fix can be applied until a  more suitable op-amp can be found.
Final possible problem is if the op-amp has too much input offset error. An offset null Pot will need to be added. Most single channel op-amps have provision for nulling.

  http://www.freeclassnotesonline.com/Ideal-741-Operational-Amplifier-Lab.php

[xavier60]

The voltage drop across D2 will ensure that the LM741's inputs are within their proper common mode range.
Early testing can be done with a higher value Rcs so that lower load current can be used for a given output.                 

[Frankentronics]

Ideally the voltage across R9 should be 0V and then increase linearly with increasing voltage drop across Rcs.

I just tested the circuit and I get 0V across the R9 resistor (reads 0V even when multi meter is in mV mode). However it stays at 0V regardless how much I increase the value of Rcs.

Since this voltage is not increasing when I increase the value of Rcs, does that mean the the first test failed?

Thanks...

[xavier60]

What value is Rcs and how much load current is flowing through it? And what voltage drop is it causing across Rcs?
Did you add the extra diode D2 in my last schematic? It will ensure that the LM741 is happy with the input common mode voltage.

[xavier60]

For example if a 1 ohm resistor is used for Rcs for testing, the load current only needs to be 100mA to be equivalent to testing with 100 milliohms and 1A of load current.
The circuit will behave in the same way in both cases.

BTW: Typical audio power amplifiers are actually like a big op-amp with inverting and non-inverting differential inputs.
http://www.ecircuitcenter.com/Circuits_Audio_Amp/Basic_Amplifier/Basic_Audio_Amplifier.htm

[David Hess]

The voltage drop across D2 will ensure that the LM741's inputs are within their proper common mode range.

The LM301A has an input common mode range which includes the positive supply so it is a little easier to use for high side current sensing.  Many old JFET input operational amplifiers also have an input common mode range which includes the positive supply.

[Frankentronics]

The LM301A has an input common mode range which includes the positive supply so it is a little easier to use for high side current sensing.

I was actually looking for an LM301 for my main circuit and could not find one. That's why I ended up using the LM471 for the main circuit. But I think I just made some progress with the current to voltage converter circuit.

Thanks...

[Frankentronics]

What value is Rcs and how much load current is flowing through it? And what voltage drop is it causing across Rcs?
Did you add the extra diode D2 in my last schematic? It will ensure that the LM741 is happy with the input common mode voltage.

Yes, I did in fact use the version with the diode, 10K resistor and 10k pot and I also tested the earlier version of the circuit, although I only included one image in my post.

In any event, I decided to pull everything off of the breadboard and rebuild from scratch and I do think that I made progress, as the circuit behaves differently. The very obvious difference is that I do get a reading in mV across the R9 resistor and when I dial the 10k pot I can make that reading go down to 0V. Also, when I increase the resistance of Rcs that reading increases for a few mV's, just as you said it should.

Since I do not have a 100m Ohm resistor I am using a 110m Ohm value there by connecting two .22 Ohm resistors in parallel. When I said that I increased the value of Rcs I just took one of these resistors out of circuit.


I also made the measurements that you asked for, so here is the data.

The unregulated incoming voltage is +24.3V.

When the power supply pot is at its max the outgoing voltage is now +22.0V

With +22.0V outgoing voltage the readings are as follows:

6.61mA current going through Rcs
0.7mv across .11 Ohm Rcs
1.4mV across .22 Ohm Rcs


With output voltage dialed down to 0V the readings are:

4.73mA current going through Rcs
0.5mv across .11 Ohm Rcs
1.0mV across .22 Ohm Rcs


I hope these readings are as expected and that the circuit behaves properly.

Please let me know what you think.

Thank you very much...


PS - Thank you also for that link. http://www.ecircuitcenter.com/Circuits_Audio_Amp/Basic_Amplifier/Basic_Audio_Amplifier.htm This is very interesting and educational for me. Although I think there is a typo there. The output buffer stage has Q4 and Q5, but the schematic has them labeled as Q11 and Q12.

[xavier60]

The behavior of the null Pot and other readings  is an indication that the circuit is working.
The purpose of the circuit is so that current can be measured. Simply put a load on the PSU's output, anything from 1 to 10 ohms.
Adjusting the output voltage will also vary the current flow. Beware of the heatsink temperature.

Also: The LM301 seem to be getting scarce and expensive. There are some reasonably priced JFET(P-CH) input TL071's on ebay if not fakes.

[xavier60]

We want to see a useful relationship between PSU output current and the voltage across R9

[Frankentronics]

I just finished some testing. I have a bunch of salvaged low Ohm resistors that I was connecting to the output load and I had a multimeter connected to measure voltage across R9. I measured voltage across R9 for each load resistor that I connected to the output.

I could get a stable reading across R9 for each load resistor that was 10Ohm or more, but the lower the values of load resistors I was using, the more unstable the readings were. I guess it was due to the heat buildup on the LM317 and on the load resistor itself.

Here are some readings I've recorded

Load Resistor - voltage across R9

1 Ohm .................390mv
1.5 Ohm ..............401mV
1.8 Ohm ..............411mV
2 Ohm ................406mV
2.2 Ohm .............431mV
3.9 Ohm ..............471mV
4 Ohm ................462mV
5 Ohm ................494mV
8 Ohm ................360mV
10 Ohm ..............306mV

I also did some readings above 10 Ohm

12 Ohm ..............261mV
36 Ohm .............106mV
50 Ohm .............77.4mv
100 Ohm ...........40.5mV
200 Ohm ...........20.7mV

Again, it was difficult to decide which reading to write down when connecting lower value load resistors because the readings on the DMM kept dropping as it was heating up. But I tried to capture the first number that showed up.

Also, I just wrote down whatever value was written on the resistor. Perhaps I should have measured each resistor first, to make sure they are on spec.

Please let me know what your thoughts are.

Thank you very much.

[xavier60]

It's not clear what the actual current is for each resistor without knowing the output voltage.
You could have just used a few of those resistors and simply varied the output voltage to change the current.
Connect it to the modified panel meter and test it. Expect the calibration to be way off for now.

You want the meter reading to accurately track the actual PSU output current.
 Because these meters don't read properly down to zero, the calibration should be done at some low current like 50mA and some higher current that can be sustained by the PSU, 500mA maybe.

Change the value of R8 to correct the high reading, this is your Gain setting. The I-ADJ Pot in the meter can be used for fine adjustment.
Adjust the Null Pot to correct the low reading, this is your Zero setting. There will be some interaction, so the adjustments will have to be done few times.
Remember to remove R9.

[Frankentronics]

You could have just used a few of those resistors and simply varied the output voltage to change the current.

Oh, I see. Will do...

Connect it to the modified panel meter and test it...

Just to be clear how to connect the modified panel meter. I need to remove R9 and connect the meter by connecting the collector of the transistor with the 100k resistor that I installed inside the meter. Is that correct?

[xavier60]

You could have just used a few of those resistors and simply varied the output voltage to change the current.

Oh, I see. Will do...

Connect it to the modified panel meter and test it...

Just to be clear how to connect the modified panel meter. I need to remove R9 and connect the meter by connecting the collector of the transistor with the 100k resistor that I installed inside the meter. Is that correct?

That's right.

[Frankentronics]

Great!

I actually already gathered some data, using a 10 Ohm load as well as a 4 Ohm load, and varying the voltage.


current across load ..... voltage across R9

0.05 A ......................... 12.8mV
0.10 A ......................... 24.1mv
0.20 A ......................... 47.0mV
0.40 A ......................... 92.1mV
0.50 A ........................ 114.7mv

This all makes sense this time.

Tomorrow I think I'll build a board of this circuit and do all the calibration straight on the board. I will add a 10k VR in series with that 10k R8 resistor so I can just dial in.

You were really helpful and I can't find the words to express my gratitude. It is clear that I could not have done this without your help.

I'll post an updated schematic and pictures of my work as I put it together.

Once I have this power supply I'll finally be able to work on some projects that I wasn't able to work on.

Thank you...

[xavier60]

Those readings look very good.
In place of R8, make plenty of provision for adjustment. At least 2 parallel resistors, one with a Trimpot in series with it.
I sometimes solder in machined IC socket pins so that I can change component values without soldering.
.
The LM358 is actually poorly suited to its function in these meters. Later on you might decide to remove it altogether. The I-ADJ Pot will likely go also, relying totally on R8 for gain adjustment.

Also: check to see if the LM741 needs the D2 diode to work properly.

[xavier60]

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

[Frankentronics]

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.