EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: fenclu on November 16, 2015, 05:49:03 pm
-
Hi
I'm designing a low-current lab bench PSU. Now, I get the basic idea why using a voltage regulator IC might not be the best idea. That's why I decided to use a voltage regulator made from discrete components.
I'm working with something like this:
(http://tangentsoft.net/elec/bitmaps/sulzer-regulator.png)
Now, I want to drive the regulator with external voltage, say from a DAC or PWM. So I basically wnt to ommit the Zener diode and the two resistors R4, R5 and just drive the non-inverting input of the opamp through an RC filter and a buffer. I bulit a simple prototype, did some simulations and it seems to work just fine with a 10uF filtering cap on the PWM. But I'm quite afraid to just call it a working circuit and move on to the next step of the design.
So, what are the basic technical aspects I have to consider when driving a voltage regulator (be it an IC regulator or a discrete one) directly from external voltage? Will it work fine without the Zener diode? Will the stability or transient response be in any way affected by using this method?
I know that Dave made a whole series on designing the ?Supply, but the whole concept of driving a regulator with an external voltage and without a reference diode still seems quite odd to me.
All help is much appreciated :)
-
The schematic you have made doesn't make much sense to me.
Why the zener on the output? Why tha crazy tantalums there?
Use a simple non-inverting amplifier. That is - an opamp and two resistor. Thats the basic principle of linear voltage amplification. Inverting input goes to the output voltage divider, noninverting input is your modulation input.
-
This isn't my schematic - it was made by a man called Sulzer. I'm not using this exact one, I just quickly googled a model. I'm using basically what you're saying - an NPN transistor with LM324 with a gain of 2.8 - a result of choosing two arbitrary resistors.
In my setup I am seeing some sawtooth-shaped fluctuations on the output - about 50mV in amplitude. Is it normal?
Please help me understand - can most of the linear voltage regulators with a feedback loop just be driven with an external voltage? Could this be that easy? :D
Please note that I am pretty much a beginner and I'm trying to learn something :D
-
Having the zener driven from the output is fairly common for fixed voltage supplies. This insures a constant current for the zener. C1 should be dumped for something a lot smaller .02 to give a faster response and get rid of R!. You may have to add more gin to the divider depending on the DAC output.
-
The schematic you have made doesn't make much sense to me.
Why the zener on the output? Why tha crazy tantalums there?
Supplying the reference element from an amplified copy of itself is a -very- old design trick to boost PSRR sky-high [well, not exactly, depends on the loop gain of your gain element].
The 4u7 across R2 are actually quite interesting. This forces the closed loop gain to just a little above unity gain very early on, so there's lots of loop gain at higher frequencies (good!). Q1 has a tad less than unity gain (depends on Ie).
-
I like voltage regulators.
A few equipment companies like hulet pakard (not sure about the spelling) have been using them for years.
But since this a thread where regulators have already been found guilty and unworthy,
here are the two relevant articles regarding the Sulzer regulators which may or may not provide enough insight, excuses or justifications for the existence of the circuit in the firstplace.
Let the games begin :popcorn:
-
I like voltage regulators.
A few equipment companies like hulet pakard (not sure about the spelling) have been using them for years.
But since this a thread where regulators have already been found guilty and unworthy,
here are the two relevant articles regarding the Sulzer regulators which may or may not provide enough insight, excuses or justifications for the existence of the circuit in the firstplace.
Let the games begin :popcorn:
Indeed, HP uses many regulator ICs in their bench power supplies. Not on the output, mind you, but to provide power the components which comprise the "real" regulated output. 8)
An HP bench supply has at least an order of magnitude ripple, noise, and drift than any regulator IC. Two orders of magnitude would probably be a closer estimate.
-
Integrated voltage regulators are designed for constant voltage output and usually a fixed current limit or a current limit set by an external shunt. Exceptions to this are rare. Most will have a (approximate) constant product SOA protection, which is not really what you want in a lab supply (significantly less output current at low output voltages). They are perfect for bias voltages. They are a poor fit for a lab supply (cf. Dave's design, which even used a specialty part and still had overall poor performance).
-
That's what I thought...so you guys would also recommend using a discrete regulator? Probably with a power NPN and a feedback loop?
Coming back to my original question - what should I look out for when driving such a regulator from an external voltage instead of using a voltage divider between the output and the non-inverting inptu of the opamp?
-
Oh my bad... The schematic in first post sure does make sense. I have looked into it only too briefly and made wrongful statement about it. Sorry.
But still, this topology won't help you make a linreg with modulation input.
What you need I have already mentioned - a noninverting linear amplifier In the principle, single opamp with two resistors. Then you can work it up: Add powerstage for more output current, add output current limiting, power limiting, all needed compensation networks, filtering caps.
You don't need any voltage reference there, as the voltage reference is the voltage you should provide by the DAC or whatever will you have there.
This you can have as a rough idea how to do such supply with a output voltage setpoint from a DAC. It also involves a simple current limiter, all basic protection circuitry. Ck is a lag compensation capacitor, Ck is compensation for the current limiter. Ck ~ some tens of nF.
The circuit is good enough for up to 30V output voltage and up to 1 Amp.
//RANT! How is that possible the forum doesn't accept JPEG as an attachment?! :palm:
-
Thanks a lot for the schematic Yansi!
And what opamp would you recommend? Probably something reasonably fast with low offset voltage? Any other considerations?
-
(http://i.imgur.com/kuypfA5.jpg)
This is basically what I'm seeing on the output. The scope is at 5mV/div. So that gives me about 10mV noise with a rough circuit setup (100uF electrolitic on the output, driven from a 1V filtered PWM from an Arduino) . What is curious to me is the triangluar shape of the waveform. Is this normal?
-
And the ripple on the amplifiers input from your "filtered PWM" is...?
-
Thanks a lot for the schematic Yansi!
And what opamp would you recommend? Probably something reasonably fast with low offset voltage? Any other considerations?
The circuit will go with almost any opamp, but two main requirements has to be fullfilled here:
1) the opamp must be a type which has "input common mode voltage range includes ground" which is a LM324, LM358 or such (and or any Rail to Rail opamp)
2) enough supply voltage headroom! If you want this supply topology to go up to 30V output, a good care must be taken that the input voltage does NOT overstress the opamp as usually their max is 36V. (These opamps are a no go for a 30V supply in this topology by the way...).
The more problematic this gets, if the input voltage is from a rectified mains transformer. You have to consider the mains voltage variation (+-10% is the quality spec of mains voltage tolerance in our area) and also you have to consider the voltage ripple on you filter capacitor, so that the regulator is able to regulate properly at the minimum mains voltage and full load but also must withstand the no-load filtercap voltage when mains +10%. (Otherwise the design is poor, like many many diy power supplies).
Yan