Vref Voltage Amplification Circuit Please?

[Mechatrommer]

this is easier thought than done. a simple thing proved to be difficult:

problem: all my opamps and ICs only can take lower power rail ie ±15V. these all are my only hope for stable control.
in that rail, there is a Vref ranging from 0.2V to 10V. this need to be translated 5X to control PSU on higher rail, ie ±60V rail.
so the final output of Vref needs to be 5X, ie = 1V - 50V.

i have high voltage transistors but cant make the arrangement. the best i can do is 1X amplification i have textbook but it seems pages to understand all that and probably i'm in a way of building myself a high voltage discrete opamp which i dont believe i'm capable of, not in this time frame. the amplified Vref will be used to control another power transistor or voltage regulator for stable PSU output 1-50V.
anyone care to help with a simplistic circuit?

[Ammar]

I'm not completely sure what you are looking to achieve, but it looks like you want to step up a voltage. One way to do this is with a boost converter. ICs that do this job are plentiful. Have a look and see if a Boost converter is what you are after. Good luck!

[MosherIV]

Hi

problem: all my opamps and ICs only can take lower power rail ie ±15V. these all are my only hope for stable control

You shlould be able to get op-amps that can take 60V single supply (not ±), then you can just use it in single supply rail mode.
Bear in mind that they might be a little difficult to source.

I was going to answer you other thread about bulding a controllable 50v psu but like you I was having difficulty finding something simple.

For a solution not involving op-amps look at this tread.
https://www.eevblog.com/forum/beginners/linear-supply-regulator-design-doesn't-seem-to-work-am-i-missing-something/
Jitter says that you can use the LM317 since the regulator 'sees' voltage within it's specification.

For the op-amp solution look at my post
https://www.eevblog.com/forum/projects/another-ac-voltage-standard/
Look at the top left corner where I am trying to design a prescision 5V supply

In both cases for the control, use a digital pot. Yes, I know others have told you that you cannot get them to work out side their supply voltage. They are wrong, most are limited to their supply voltage but Analogue Devices do make digipots that work outside their supply voltage. Check carefully if the range they can do will work in your circumstances, 50V is on the high side.

[Mechatrommer]

One way to do this is with a boost converter.

i'm aware of boost topology but i guess it introduce more noise than necessary? i'm looking for lower noise linear solution if there is any, not that i need super stable uV spec, but at least not the SMPS noise spikes.

You shlould be able to get op-amps that can take 60V single supply (not ±)

yes searching for high voltage opamp took me less than few seconds, the only problem is, i will seldomly design high voltage like this so i dont think i want to collect part just for this particular application. simply put, i dont have the part right now that can even take 60V single supply (V-GND). at most is 30-38V supply like LM324, LM071 and uA741.

For a solution not involving op-amps look at this tread.
https://www.eevblog.com/forum/beginners/linear-supply-regulator-design-doesn't-seem-to-work-am-i-missing-something/
Jitter says that you can use the LM317 since the regulator 'sees' voltage within it's specification.

i'm building psu like that, dual track with pre-regulation, except up to 50V. and i'm going to use the exact topology like that (Vin-Vout) only within few volts on the final step using 317 if it feasible, but thats not what the problem here. the problem is producing (1-50V) from 60V supply. 59V differential between LM317/337 input-output is not going to cut it. think about the output of OPA27 (A2) in that diagram, there lies my problem. even OPA27 cant take 60V single supply, its max input is 44V. there is another OPA445 that can do exactly what i want, but thats beyond my reach. i will reach it if there is no simpler solution.

For the op-amp solution look at my post
https://www.eevblog.com/forum/projects/another-ac-voltage-standard/
Look at the top left corner where I am trying to design a prescision 5V supply

not in any way similar. yours only goes 12Vmax, mine is 60V and OP07 can only take 36V single supply.

In both cases for the control, use a digital pot. Yes, I know others have told you that you cannot get them to work out side their supply voltage. They are wrong, most are limited to their supply voltage but Analogue Devices do make digipots that work outside their supply voltage. Check carefully if the range they can do will work in your circumstances, 50V is on the high side.

i'm using mechanical pots, the simplistic solution i have no problem with, in fact i wont try to resolve to more complicated digital ic solution where possible. thanks for your reply anyway.

[Mechatrommer]

fwiw, my trimpot comes from 15V rail feed to few low volt (jellybeans) opamps and comparators for pre-regulation switching, that part is done. now the same Vref from 15V rail (buffered by an 15V opamp) needs to be amplified 5 times, for linear front end regulation from 60V rail. 5 - 55V preregulated input rail to be exact. but the 1-50V Vref need to be flat stable throughout the regulation, hence needing its input from another stable 60V low power rail.

[danadak]

Some approaches -

http://www.electronics-lab.com/make-an-op-amp-high-voltage-output-circuit/

http://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Broskie_John/HV_Regulator_OP27_IRF820.pdf


Regards, Dana.

[bktemp]

You don't need a 60V opamp, you need a level shifter for driving the preregulator.
It depends on what you are trying do do, but it could be as simple as single transistor + a some resistors.

[Mechatrommer]

pre-regulated part is done and tested beautifully. then i'm stuck here at the final high voltage linear regulation stage.

[bktemp]

It is probably much easier to find a solution if you show us the schematic of your preregulator.

[MosherIV]

yes searching for high voltage opamp took me less than few seconds, the only problem is, i will seldomly design high voltage like this so i dont think i want to collect part just for this particular application. simply put, i dont have the part right now that can even take 60V single supply (V-GND). at most is 30-38V supply like LM324, LM071 and uA741.

Fair point, collecting obscure parts can be problematic.
What are the parameters for the 50V psu?
If the LM324, LM071 or 741 have the single rail voltage, how much voltage error and niose can your application withstand? (Offset and niose are the main dissadvantages of these general purpose op-amps but the offset error is in the sub milivolt region, cannot remeber what the noise parameter is. Are they good enough?)

For the 55V single rail psu, I was refering to the reference design at the bottom of the page, just replace the 4.7K and 27K for suitable values for your application, or a pot in your case. Bear in mind, in this configuration it will not be able to go down to 0V or 1.25V because the reference pin is now well above 0V

i'm building psu like that, dual track with pre-regulation, except up to 50V. and i'm going to use the exact topology like that (Vin-Vout) only within few volts on the final step using 317 if it feasible, but thats not what the problem here. the problem is producing (1-50V) from 60V supply 

Presumably, that is a different project. I do not think you will be able to get the LM317/337 to do 0 to 50V. You do not need the op-amp in the circuit to getdown to 0v. You just pull the reference of the LM317 to -1.25V and the LM337 to 1.25V and use a dual pot to control both for a far simpler circuit, the dissadvantage of doing it this way is that there will be a small difference between the + and - rails but for most simple applications, this is not a problem. This will give a simple 0 to 25V dual supply.
However, I would suggest investing in a good bench supply. It will give you gaurenteed regulation and a current limit (very important when trying circuits out). Your need for a 50V rail will mean a little more expense but interms of better protection, it will be worth it. Designing and building your own will involve significant time and effort (and parts!). I am not sure if Dual Tracking Bench PSUs go up to 50V

[Zero999]

Add some other transistors to boost the voltage.

Here's an example using the TL431.

[Mechatrommer]

Some approaches -
http://www.electronics-lab.com/make-an-op-amp-high-voltage-output-circuit/
http://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Broskie_John/HV_Regulator_OP27_IRF820.pdf
Regards, Dana.

seems like what i want, the need for a low volt opamp wont hurt, but i still try to get my head around it. thanks for the idea, that made me search "low volt opamp to high volt output" in google and i found this. this looks pretty only few transistors. need to study this, thanks:

It is probably much easier to find a solution if you show us the schematic of your preregulator.

i think not. these are modular design i tried to part away the pre-regulation part from linear regulation part, the only matter come into contact with linear regulator is the Vref. but FWIW, here's the schematics so far (attached), critics and comments are welcomed, but utmost important is the Vref part (Vrc+ and Vrc- middle upper part in the sch)...

What are the parameters for the 50V psu?

50W, thats the crude test i made few days back that the transformer can spit without significant sag. since this is preregulated somewhere 3-5V diff on the 317 power pins (or power bjt, depending on the final decision that is yet to be made) output, i'm thinking probably it can do 50A at 1V output, but this is something to be discovered. at 50V its @1A. i may lower this spec if i find its unacceptable. noise and offset is not my main concern since i'm suck arse at em, offset can be compensated using some trimpots adjustment in control stage (Vref) but noise, well... the best that i can achieved

@Hero. you always come to rescue thanks alot. if you look at the sch, you should notice the mosfet's driver at +ve rail came from your suggestion, i just have to figure out the mirror of it for -ve rail many many replies my brain is slow, thanks all.

[MosherIV]

50W, thats the crude test i made few days back that the transformer can spit without significant sag. since this is preregulated somewhere 3-5V diff on the 317 power pins (or power bjt, depending on the final decision that is yet to be made) output, i'm thinking probably it can do 50A at 1V output, but this is something to be discovered. at 50V its @1A. i may lower this spec if i find its unacceptable. noise and offset is not my main concern since i'm suck arse at em, offset can be compensated using some trimpots adjustment in control stage (Vref) but noise, well... the best that i can achieved 

A 50VA(50W) transformer will nowadays not be able give 50VA, they skimp on the magenetic which means as you get to the top limit, the magnetic core saturates and you cannot get the rated output. You need to over rate the spec of the taransformer to get the performance you want.

Noise and offset parameters are what will determine how the psu will be and whether the general purpose opamps are suitable.
Is 0.01V offset or error good enough on the Voltgage output?
Is 0.005V noise ontop of the output good enough?
If the asnwer is yes to both these, then the general purpose opamps should be good enough.

[bktemp]

It is probably much easier to find a solution if you show us the schematic of your preregulator.

i think not. these are modular design i tried to part away the pre-regulation part from linear regulation part, the only matter come into contact with linear regulator is the Vref. but FWIW, here's the schematics so far (attached), critics and comments are welcomed, but utmost important is the Vref part (Vrc+ and Vrc- middle upper part in the sch)...

Can you explain what you are trying to do? Maybe I miss something, but the circuit looks weird.
I can see the stepdown converter for both positive and negative voltage rails, but the drive input is connected to a comparator comparing the preregulated output voltage to some other voltage. I don't see an oscillator or other components for generating the PWM signal.
And I don't see where you need the 0-50V signal. You already have a highside mosfet driver with a level shifter. By the way: You mosfet driver needs some improvement, because it drives the mosfet with 60V on the gate! Most mosfets only tolerate 20V.

[bktemp]

A 50VA(50W) transformer will nowadays not be able give 50VA, they skimp on the magenetic which means as you get to the top limit, the magnetic core saturates and you cannot get the rated output. You need to over rate the spec of the taransformer to get the performance you want.

If a 50VA transformer is not able to supply 50VA into a resistive load, it is simply not a 50VA transformer.
Saturation has nothing to do with output current, it only depends on the voltage applied to the winding: The core gets only saturated if the input voltage is too high (for a fixed frequency with no dc bias). If you draw more current, there will be a larger voltage drop on the primary, reducing the voltage seen by the inductive part, so it satures less than at no load condition.
If you add a rectifier the transformer needs to be overrated, because of the high peak currents generated when rectifying the voltage. Those peaks generate a higher rms current than the avarage output current, so you have higher losses in the transformer compared to a resistive load. So for 50V 1A dc you need a transformer rated for at least 50V 75VA. 50Vac will give probably 75Vdc at no load.

[Mechatrommer]

Noise and offset parameters are what will determine how the psu will be and whether the general purpose opamps are suitable.
Is 0.01V offset or error good enough on the Voltgage output?
Is 0.005V noise ontop of the output good enough?
If the asnwer is yes to both these, then the general purpose opamps should be good enough.

thanks for your concern, but you dont get me when i said i'm a suck arse on this. so here it is:
0.1V offset is good enough.
0.05V noise is good enough. 0.01V is better.

Maybe I miss something, but the circuit looks weird.
Can you explain what you are trying to do?

what i'm trying to do (and what you should help me) does not yet exist in the drawing. its a voltage reference for the next "linear regulator circuit", everybody who watch dave's video knows about it.

I can see the stepdown converter for both positive and negative voltage rails, but the drive input is connected to a comparator comparing the preregulated output voltage to some other voltage. I don't see an oscillator or other components for generating the PWM signal.

this is "novel" "PWM on demand" SMPS pre-regulator the PWM is generated by the 393 whenever the preregulated voltage goes below Vpc+ (and Vpc- for -ve rail). dont mind too much on it, i'll take full responsibility on this side.

And I don't see where you need the 0-50V signal.

it goes from Vrc+ and Vrc- to way beyond the boundary of the schematics drawing, page 2, does not yet exist.

You already have a highside mosfet driver with a level shifter. By the way: You mosfet driver needs some improvement, because it drives the mosfet with 60V on the gate! Most mosfets only tolerate 20V.

you are right! i missed that thanks. i will ammend the drawing, maybe i'll shift the driver side to 15V rail. the note there: whats not shown is, i tested with 30V rail on IRF4905 on breadboard, my main concern was about the bjt driver (Hero999 setup). but i guess i've abused it with 30Vgs for no good reason, i think its still alive i hope. thanks.

A 50VA(50W) transformer will nowadays not be able give 50VA, they skimp on the magenetic which means as you get to the top limit, the magnetic core saturates and you cannot get the rated output. You need to over rate the spec of the taransformer to get the performance you want.

If a 50VA transformer is not able to supply 50VA into a resistive load, it is simply not a 50VA transformer.
Saturation has nothing to do with output current, it only depends on the voltage applied to the winding: The core gets only saturated if the input voltage is too high (for a fixed frequency with no dc bias). If you draw more current, there will be a larger voltage drop on the primary, reducing the voltage seen by the inductive part, so it satures less than at no load condition.
If you add a rectifier the transformer needs to be overrated, because of the high peak currents generated when rectifying the voltage. Those peaks generate a higher rms current than the avarage output current, so you have higher losses in the transformer compared to a resistive load. So for 50V 1A dc you need a transformer rated for at least 50V 75VA. 50Vac will give probably 75Vdc at no load.

thanks for the info. my transformer put ±63.2V on capacitors after rectification. but as i said, i will change the spec if the transformer or PSU performance cant take it, i'm no transformer expert anyway, only what i've tested empirically.

[bktemp]

I can see the stepdown converter for both positive and negative voltage rails, but the drive input is connected to a comparator comparing the preregulated output voltage to some other voltage. I don't see an oscillator or other components for generating the PWM signal.

this is "novel" "PWM on demand" SMPS pre-regulator the PWM is generated by the 393 whenever the preregulated voltage goes below Vpc+ (and Vpc- for -ve rail). dont mind too much on it, i'll take full responsibility on this side.

It probably will regulate the output voltage to the desired value somehow, but it will have a bad efficiency because the operating frequency will be random. It could be 100Hz, but it could also oscillate somewhere at >100kHz. And it could also be worse: Because there is no hysteresis, it could even operate in linear mode. If it operates at a too low frequency, the peak current will be high and saturate the inductor resulting in high losses in both inductor and mosfet. To sum it up: It is a bad design.
If you don't want to build/use a complete PWM controller I would at least add some hysteresis to the comparator (maybe a small capacitor, that will make it oscillate because it adds a time dependent hysteresis).

And I don't see where you need the 0-50V signal.

it goes from Vrc+ and Vrc- to way beyond the boundary of the schematics drawing, page 2, does not yet exist.

Vrc is the set voltage controlling the Adj terminal of the LM317 regulator?

A LM317 does not work at 50V: Even if it sees only a small voltage difference across it during normal operation, if the output gets shorted it will see the full voltage from the preregulator output caps and fail much faster the preregulator can respond.

You already have a highside mosfet driver with a level shifter. By the way: You mosfet driver needs some improvement, because it drives the mosfet with 60V on the gate! Most mosfets only tolerate 20V.

you are right! i missed that thanks. i will ammend the drawing, maybe i'll shift the driver side to 15V rail. the note there: whats not shown is, i tested with 30V rail on IRF4905 on breadboard, my main concern was about the bjt driver (Hero999 setup). but i guess i've abused it with 30Vgs for no good reason, i think its still alive i hope. thanks.

It is easy to improve the mosfet driver. See attached schematic.
Q1 acts as a current source, supplying (Vbase-0.7V)/R2. Let's say you adjust R2 to 5mA to keep the power dissipation at 60V at <0.25W, then 3k for R1 will generate a gate voltage of 15V. Because 5mA are much too low for driving the mosfet directly, Q2/Q3 act as a voltage buffer increasing the output drive. At 60V all transistors need to be rated for at least 70V, something like BD139/BD140 should work for Q2/Q3. You could also add a resistor in the collector line from Q3 to distribute the power dissipation when it turns the mosfet on, because all the gate current is taken from the 60V, generating a large power dissipation for a short time.

[Mechatrommer]

It probably will regulate the output voltage to the desired value somehow, but it will have a bad efficiency because the operating frequency will be random. It could be 100Hz, but it could also oscillate somewhere at >100kHz. And it could also be worse: Because there is no hysteresis, it could even operate in linear mode. If it operates at a too low frequency, the peak current will be high and saturate the inductor resulting in high losses in both inductor and mosfet. To sum it up: It is a bad design.

yeah, the inductor in sch is only when an inductor is needed, to stabilize and experimentation a bit. i just put there so i have place for an inductor to be soldered on, in reality its just a dead short. so not a true buck convertor. the preregulated voltage may oscillate badly i only have to make sure it didnt go below drop out voltage of a linear regulator in the next stage (through adjustable Vpo offset). so there it is of me being suck arse tried to simplify my circuit as much as my head can get around to. i have MC34063 (SMPS DC/DC Converter Boost Buck Regulator) i thought this is what you were talking about PWM controller in earlier post, but i dont understand it fully and i think i'm going to use those on more demanding PSU spec later. this PSU i'm building is just "as long as it can power up my test project" kinda thing. and since the power comes from the mains, efficiency is not the main concern.

If you don't want to build/use a complete PWM controller I would at least add some hysteresis to the comparator (maybe a small capacitor, that will make it oscillate because it adds a time dependent hysteresis).

good idea! thanks.

Vrc is the set voltage controlling the Adj terminal of the LM317 regulator?
A LM317 does not work at 50V: Even if it sees only a small voltage difference across it during normal operation, if the output gets shorted it will see the full voltage from the preregulator output caps and fail much faster the preregulator can respond.

good point, i'll use crude power bjt setup then.

It is easy to improve the mosfet driver. See attached schematic.
Q1 acts as a current source, supplying (Vbase-0.7V)/R2. Let's say you adjust R2 to 5mA to keep the power dissipation at 60V at <0.25W, then 3k for R1 will generate a gate voltage of 15V. Because 5mA are much too low for driving the mosfet directly, Q2/Q3 act as a voltage buffer increasing the output drive. At 60V all transistors need to be rated for at least 70V, something like BD139/BD140 should work for Q2/Q3. You could also add a resistor in the collector line from Q3 to distribute the power dissipation when it turns the mosfet on, because all the gate current is taken from the 60V, generating a large power dissipation for a short time.

thanks! i'll study that.

[Zero999]

Some approaches -
http://www.electronics-lab.com/make-an-op-amp-high-voltage-output-circuit/
http://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Broskie_John/HV_Regulator_OP27_IRF820.pdf
Regards, Dana.

seems like what i want, the need for a low volt opamp wont hurt, but i still try to get my head around it. thanks for the idea, that made me search "low volt opamp to high volt output" in google and i found this. this looks pretty only few transistors. need to study this, thanks:

I see what you're trying to do now.

How much current does it need to control?

That circuit will work but it has a lot of gain so it's likely to oscillated. Adding more negative feedback inside the buffer loop aught to stabilise it.

[Mechatrommer]

How much current does it need to control?

no idea maybe few tens of millivolt the lower the better. this 1-50V will be feed to LM317 (adj) pin or to the base of power darlington on final linear output. but this decision will be finalized along the design process. thats me the non professional guy

That circuit will work but it has a lot of gain so it's likely to oscillated. Adding more negative feedback inside the buffer loop aught to stabilise it.

yup i dont understand why there is a cap on the feedback line, that makes DC gain open loop, i removed the cap already and it worked in sim this way. or i'll probably add another cap from opamp inv input to gnd to attenuate oscillation (hi freq) gain. in your circuit you've added R1, not sure what it does but adding that footprint into the schematic wont hurt, thanks.

[Zero999]

yup i dont understand why there is a cap on the feedback line, that makes DC gain open loop, i removed the cap already and it worked in sim this way. or i'll probably add another cap from opamp inv input to gnd to attenuate oscillation (hi freq) gain.

The capacitor there was to slow down U1. I'd suspect connecting it from the inverting input to 0V might make it worse though.

in your circuit you've added R1, not sure what it does but adding that footprint into the schematic wont hurt, thanks.

Q2's emitter is a place to insert some negative feedback to reduce the gain.

When the voltage on Q2's base rises, it turns on, turning on Q1. When the voltage on Q2's emitter rises, it turns off, causing Q1 to turn off more. Connecting a potential divider from the output to Q2's emitter increases the emitter voltage, as the output voltage increases, thus turning off Q2 and Q1: i.e. it adds negative feedback. The gain of the output stage is roughly equal to 1+ R1/R2 or 7.67 in this case. As long as the gain of the output stage is under 1+R3/R4 then the circuit should be stable.

[David Hess]

This circuit can be simplified.

Drive the base of Q1 through a zener diode connected to the + supply input of the operational amplifier.  The operational amplifier's output drives ground through a resistor and the collector of Q1 drives the operational amplifier's output through another resistor fixing the voltage gain of Q1.  Global feedback from the output back to the non-inverting input of the operational amplifier sets overall gain.

Effectively the output transistor of the operational amplifier replaces Q2 and the output of the operational amplifier becomes a third high bandwidth current input.  Bipolar designs are more common as shown in figure 2 on page 2 of Linear Technology application note 18.  There is a better example in their application notes but I do not remember exactly where.

Of course you will not want to do this with with a dual operational amplifier.

[Zero999]

This circuit can be simplified.

Drive the base of Q1 through a zener diode connected to the + supply input of the operational amplifier.  The operational amplifier's output drives ground through a resistor and the collector of Q1 drives the operational amplifier's output through another resistor fixing the voltage gain of Q1.  Global feedback from the output back to the non-inverting input of the operational amplifier sets overall gain.

Effectively the output transistor of the operational amplifier replaces Q2 and the output of the operational amplifier becomes a third high bandwidth current input.  Bipolar designs are more common as shown in figure 2 on page 2 of Linear Technology application note 18.  There is a better example in their application notes but I do not remember exactly where.

Of course you will not want to do this with with a dual operational amplifier.

Yes, that will work, as long as the 60V rail is stable enough to be within the op-amp's ratings minus the zener drop. It may even work with a linear regulator, such as the LM7812.

Stability may remain to be an issue though, so the gain could be reduced by keeping R1 & R2 in place and connecting the op-amp's output to where Q2's emitter currently is.

[David Hess]

Yes, that will work, as long as the 60V rail is stable enough to be within the op-amp's ratings minus the zener drop. It may even work with a linear regulator, such as the LM7812.

I am not sure what you mean by a 7812 linear regulator.

The voltage range will be limited by the operational amplifier's supply range and if that is a problem, then a cascode transistor can be added to the positive supply terminal.

Stability may remain to be an issue though, so the gain could be reduced by keeping R1 & R2 in place and connecting the op-amp's output to where Q2's emitter currently is.

My description could have been better but this is what I described and the circuit reference I included shows it.  The frequency compensation of the circuit should be tested and adjusted with the expected output load.

There are fully discrete voltage gain output stages which are not much more complicated but they perform worse although they are easier to understand.  An alternative lower noise design uses a zener level shift so the operational amplifier can drive an emitter/source follower directly but this limits the output voltage range to the range of the operational amplifier so is not suitable if the output needs to extend to ground.

[Zero999]

I didn't see any circuit reference.

I take it you mean like this? Yes, that'll work.


What I meant was the zener might be able to be replaced by the LM7812 but it could be unstable.

No, forget that, the circuit is powered from 60V and the LM7812 won't like that. The LM317 set to 30V would work (as it's the Vin-Vout that matters) but as I said before, stability could be more of an issue.

Of course you will not want to do this with with a dual operational amplifier.

Well, you could just not use the other op-amp in the dual package.

However, you could use it, just as long as you can ensure the current sourced by the output isn't enough to switch on Q1.