Modifying the bench PSU to run at 38V

[alecjcook]

Hi,
   I have been watching Dave's video blog about building yourself a bench power supply. I have avidly followed his design and wish to alter it a little but I am struggling a little, so I thought maybe someone here could help.
Video Blog Entries "#221, #222, #224..."

This PSU seems great for the 10V as spectified, but the LM3080 is rated up to 38V, I thought it would be nice to tweek the design. I have wound up the voltage to 24V (not even 83 yet) and when my current limiter kicks in, I start dumping 8Amps through my Current Pass Transistor. 

I obviously need more control over the current, but could someone tell me what in my circuit I have to tweek (and why) to better control this please?
Many thanks, Alec

Note: this is Dave's design, with the voltage reference removed, the PSU upped to 24V and the value of R2 tweaked so that 1V is still fed into the I.ADJUST op amp.

[fcb]

Am I being thick?  I can't see an LM3080 (LT3080 typo?).

Or your pass transistor (LM317...?)

Or any mechanism to dump current (other than the load)??

[lapm]

IS that schematic correct? Seems to me like current sense resistor is in wrong place? Or is it missing a wire?

[alecjcook]

Im sorry, I forgot to mention that National Instruments Multisim doesn't have an LM3080 on offer, so I am substituting it with a LM317.
My transformer is only going to deliver 50VA, so tapping of at 48V will not leave me much current to dump. Is this OK, or will it do something weird like put too much strain on the transformer/circuitry?

[alxnik]

1. You might want to connect the trace between R9 and R1 to Vcc, or else the output will be rated in mA if not uA 

2. I think you mean LT3080? LM3080 is a different part

3. LT3080 goes up to 36 volts, so 38 will probably fry it. I would stay around 30V to be on the safe side. Who needs 38V after all??

4. Feeding the opamps input from the supply rail, effectively renders the whole regulator concept pointless. The same ripple you have at input, you will have at output. The voltage reference is essential!

5. Not sure what the bulbs are there for. If you want them for load testing, I would suggest a linear resistor.

[alecjcook]

Hi,
OK thanks, that answered none of my questions. Yes I mean LT not LM. Yes I mean 36 not 38. No I am not feeding the input of the op amps with the supply rails only the supply of the op amps. Yes I am using bulbs for a reason because it's a nice visual indicator (if they go over voltage of there is too much current they look like they blow up  so I can tell that my circuit is doing the ting that I want)

So. This circuit is the one built by dave on his blog, the Tract between R9 and R1 to Vcc is not required because mA accuracy is everything i need for this project and not more. As I said before, this circuit works fine when supplied with 7V, it works and simulates perfectly. Now I have upped the supply and tweaked the value of R16 to allow the op amp to be being fed with a value closer to the 1V per V which my circuit requires, unfortunately the current limiter has stopped working. Does anyone know why this is?
Many thanks, Alec

PS here is a new schematic to keep alxnik happy

[alxnik]

Hi,
OK thanks, that answered none of my questions. Yes I mean LT not LM. Yes I mean 36 not 38. No I am not feeding the input of the op amps with the supply rails only the supply of the op amps. Yes I am using bulbs for a reason because it's a nice visual indicator (if they go over voltage of there is too much current they look like they blow up  so I can tell that my circuit is doing the ting that I want)

So. This circuit is the one built by dave on his blog, the Tract between R9 and R1 to Vcc is not required because mA accuracy is everything i need for this project and not more. As I said before, this circuit works fine when supplied with 7V, it works and simulates perfectly. Now I have upped the supply and tweaked the value of R16 to allow the op amp to be being fed with a value closer to the 1V per V which my circuit requires, unfortunately the current limiter has stopped working. Does anyone know why this is?
Many thanks, Alec

PS here is a new schematic to keep alxnik happy

I don't think you understood me. The circuit you have here is not identical to dave's, and it's quite wrong, actually it works by pure luck.

1. You have no power path between Vcc and the regulator. Dave's does. If you get any mA (not accuracy, total power output) it will be from the opamp outputs. If you only need this, you are better off with a battery.

2. You removed the voltage reference from after R5. This effectively makes the whole circuit a voltage divider and not a regulator. I am guessing, the reason you did this was because there was not enough reference voltage to satisify your increase in Vcc, so you just stripped it off. Actually in order to do that properly, you have to keep the Vref in and play around with the gain network (R16/R17 for voltage).

If you need mA accuracy, why don't you just put an LM317 with a couple of resistors and be done with it?

[fcb]

Can I suggest that you look at the 10001 ohm (!) resistance feeding the regulator as a starting point?

Also, I would be tempted to post the precise specification of what you want to achieve, and perhaps explain the reasons you want to achieve it, I have a feeling that this will get you to a working result more quickly than trying to debug a fairly complex and somewhat over-specified design.

[alecjcook]

Oh really? Damn, just when I thought I was starting to get to grips with this thing.
I am looking to design a Bench PSU, I need it to deliver 24V (with maybe a little headroom i picked 36V because of the max rating of the LT3080) and about an amp.
It needs a voltage control and a current limiter (and a current limit indicator LED). I'd like it to be able to display the current and voltage before I switch the output on. I have already bought the transformer so I guess it needs to work around that (50VA: 240 In and tap offs at 12V 24V 48V and 110V). I realise that this transformer at 48V will only give me about 1 amp but that's fine too.

I won't be able to use veroboard with that amount of current so I was thinking about etching my own PCB once I have a design finalised.

After trawling though many designs, I found that dave's circuit made the most sense to me, but right at the end of the blog descriptions the circuit seemed to get a whole load more complicated without description.

I will keep playing with some of alxniks suggestions, but if fcb or anyone else could recommend a more sensible design I would be please to play around with it

Many thanks, Alec

[alxnik]

first of all, if you just want a power supply fast and simple, I would suggest that you replace the LT3080 with an LM317. It is cheaper, can be easily sourced and you can find endless circuits around the net with constant current and constant voltage.

Actually, if you look at eevblog #221 @ ~4:30 you will see the classic way to do a CC/CV with 2 lm317. Although I don't like the idea of 1amp going through a pot, it's a starting point...

Considering veroboard, you can easily build it on vero. Just buy one with points, not tracks and do point to point connections with some cable that can handle 1Amp.

[alecjcook]

OK, I'm going to adjust my spec a little. I have been doing a bit of reading and realised that my Transformer will probably be kicking out 24V RMS, not 24V DC. Meaning I will be able to get almost 34V out of it @ 2 Amps. and I really don't like the idea of putting 2 Amps though my lovely ten wind 10k pot :/

[ahnuts72]

You need to connect the top of r1 to somewhere around c5 and put the voltage reference back where Dave had it just pick a 5 volt one then change r17 and r16 to have a gain of 5 like r17=4k and r16=1k will give you 25 volts output just set the gain for whatever maximum output you want just dont make the resistors to small.
And connect the current control pot to the 5 volt reference as well that way you dont have the full supply voltage going through the pots each will see no more than 5 volts.
The resistor feeding the current control pot will set the maximum current.

Sent from my CM11 Nook HD+

[fcb]

OK, I'm going to adjust my spec a little. I have been doing a bit of reading and realised that my Transformer will probably be kicking out 24V RMS, not 24V DC. Meaning I will be able to get almost 34V out of it @ 2 Amps. and I really don't like the idea of putting 2 Amps though my lovely ten wind 10k pot :/

I think you need a little more help understanding things...

1. If you trying to feed 2A through a 10K pot (as a potential divider) you need 20,000 volts....
2. If you are feeding 2A between the wiper and one end of the 10K pot (using the pot as a variable resistor), it's an unconventional methodology, and highly likely to smoke the pot at low ohms settings.

[alecjcook]

So is an op-amp driven design not going to a little more sensible?

[alxnik]

So is an op-amp driven design not going to a little more sensible?

opamp driven design is much more configurable to the exact specs you want but much harder to correctly implement. Keep in mind that an LM317 or LT3080 are essentially opamp based regulators in a single package.

Look around (or try for yourself) to have a fixed sense resistor (for 2amps max, I would suggest 0.5ohm or lower) and then add the pot on the feedback loop of the LM317 (sense pin). Now you might have a problem that the LM317 can't handle 2 amps, but shoot for the 1.5amp that it can handle and you can parallel 2 LMs later.

[alecjcook]

OK, so going back to my original problem. I believe that the problem is that I need a component to replace the AD741 in dave's circuit for one that will work at a 36 or 38V supply voltage and operate well enough to server as the differential amp in the circuit. I have to simulate this circuit with a LT1013, but the circuit breaks down when I attempt to use this circuit as a differential op amp. Any ideas on a replacement component?

[fcb]

So find a spice model for the actual opamp you want to use, or failing that build the circuit.

[alecjcook]

ok, so my question now is, how can i just 'build it'. the only way i know how to prototype stuff is on breadboard, and if i use breadboard and stick 2 amps through it im going to vaporise it

[alxnik]

ok, so my question now is, how can i just 'build it'. the only way i know how to prototype stuff is on breadboard, and if i use breadboard and stick 2 amps through it im going to vaporise it

Build on breadboard. Don't pass 2 amps through it. Does it work? If yes, build on vero with proper cabling.

FWIW, I pass 2 amps through breadboard all the time. However I power it for a few seconds just to capture the relevant data on the oscilloscope and then turn it off. Not very accurate since it has bigger resistance on the tracks, but good for ballpark data.