>=1MHz SMU high-current output stage?

[David_]

Hello.

I have an ambition to create a Source Measure Unit(SMU) sort of device since I can't afford a real one, it don't need to be even close in specification to commercial SMU's and as a first version I thought I would design it to fit the output range of a DC-DC converter that I am planning to make in order to analyse it's operation.

I was thinking of something like +-5V @ 1A, but I quickly ran into a wall.

How would I go about buffering a DAC output signal so hat the final output could be loaded with 1A?

Is that unrealistic if I want to be able to put out waveforms up to 1MHz?

Any advice is greatly appreciated.
Cheers

[T3sl4co1l]

Some audio amps will do that range of V, I and F.

It might be a custom one, not an off the shelf op-amp style block.  But getting that kind of BW in a traditional discrete audio circuit is no biggie.

Measuring voltage and current independently will be the bigger challenge

Tim

[rfeecs]

High power op amp?  Here's one:
http://www.ti.com/lit/ds/symlink/opa564.pdf

[alm]

SMUs are mostly DC devices. They can do sweeps and pulses, but these are usually in the order of 1 ms or 100 µs pulse width. Hardly something that needs 1 MHz bandwidth. Keep in mind that it generally operates in a source-settle-measure cycle, so any advantage in improving output bandwidth would be swamped by the settling delay and sample time.

What you are building sounds more like a DC-coupled amplifier, look at the Kepco BOP or EMI/Lambda BOS/S units for something that might be similar to that (although they only go up to a couple of dozen kHz). These are not precision devices (unlike commercial SMUs), and will not provide fast or accurate readback (basically good enough for the front panel panel meters). These are useful units, so trying to clone them can be useful, but SMUs they are not. The big challenge will probably be getting them stable. Even the commercial units are not very stable in reactive loads.

If you want to build an SMU, my suggestion is to pick a commercial SMU (like the Keithley 2400 or Keithley 236 if you want full schematics), then eliminate the ranges that make it tricky (e.g. anything over 20 V, below 1 µA) and multiply the tolerances by a factor of ten or more. That might be a target to try to achieve. No point in trying to exceed the commercial units unless you have specific requirements.

[T3sl4co1l]

SMUs are mostly DC devices. They can do sweeps and pulses, but these are usually in the order of 1 ms or 100 µs pulse width. Hardly something that needs 1 MHz bandwidth. Keep in mind that it generally operates in a source-settle-measure cycle, so any advantage in improving output bandwidth would be swamped by the settling delay and sample time.

So what if you wanted a 1us settling time, or less?

I could only speculate what the OP is after, but assuming they're after something that's simply faster than the above applications -- that would indeed be the ticket to pulling it off.

Tim

[David Hess]

I know from previous experience that a discrete current feedback amplifier in that voltage and current range can achieve a bandwidth of at least 1/4 the output transistor Ft so 1 MHz of bandwidth is not difficult.

If I just wanted a voltage output which is all that you mention, then I would use some variation of the circuits shown in figure 2 of Linear Technology application note 18 or Figure 30 of National Semiconductor application note 29 where the supply pins of the operational amplifier are used as outputs to drive a discrete booster stage with local current feedback applied to the output of the operational amplifier.

[Kleinstein]

In principle 1 MHz Bandwidth is possible. However for a low impedance output, parasitic inductance can cause quite some trouble. So one might need to compromise between output impedance and bandwidth.

I would first start with a slower version to get a rough idea. Like with lab power supplies, a second less powerful unit is still useful.

[David_]

Thank you for all answers.

To be honest I am not really sure about what it is I really want, my interest came about like this.

I discovered not too long ago what a SMU is and it is exactly the sort of device that tickles my fancy, in this case it would be a device with which I could ether(or both) supply the power to a power supply or load the power supply with a load while exposing the power supply to ether input or output transients or other occurrences that challenges the power supplies regulation characteristics.

At the same time I have been interested in boosting the current capability of a function generator output, so that I could as an example create B-H loop plots characterising inductor cores. Online there are a couple of circuits for that purpose which I have used but it could be done much more simply and also characterise higher-power cores if I could buffer the output of my function generator to be able to supply the generated waveforms at currents in the 1A and above range.

Although these two things are quite different they do present with with the very similar situation of wanting to be able to freely generate waveforms and have those waveforms drive large loads(high currents).

I have heard about that audio amplifier option before but I haven't manage to find a suitable amplifier and I find that it is hard t search for such amplifiers(perhaps because I am so unknowledgeable about audio stuff) but I will look up what has been suggested in previous answers.

Most of my ambitions when it comes to what sort of circuits I want to use and design relates to circuits well outside my knowledge range, take a SMU for example, although I do know what constitutes a signal chain for collecting samples ending up as voltage or current values on a processor of some sort I really do not know anything about how I would actually use a SMU to analyse a DC-DC converters characteristics for example. And since I cannot afford a commercial SMU I feel as I need to design something that is capable of performing such tasks(with a much much lesser degree of accuracy and resolution) so that I can use that and explore what I can do with it and learn from that.

I am pretty confident that I can(in large part due to the increasable resource that internet forums such as this presents) to design something like that, and I think I can manage that quite well on my own if we neglect that very important part of enabling a DAC to produce waveforms that can be supplied while supplying higher currents at whatever frequency.

I will definitely look into the idea of using a complete SMU schematic and simplify it(although I have tried to make sense out of such schematics before but now that I think about it I simply got side-tracked and forgot about that option. That happens to me all the time due to ADD), that would be a big help by amongst other things present already defined parameters because I could easily set out to design something with features simply not needed sine the goal and application of the to be designed device is so loose.

[alm]

I would try to find commercial examples of what you do, to give some guidance in what you might need (let their engineers figure out the requirements for you). There are some impedance analyzers that may be able to do something like that. The HP 3245A universal source is another that comes to mind, but was fairly niche (I believe the catalog page mentions testing modems) and was discontinued a long time ago. I already mentioned the Kepco BOP and Lambda/EMI BOS/S four-quadrant DC-coupled power amplifiers (there are schematics for both, but they may or may not be public). You should be at the very least be able to find block diagrams for all of them.

Maybe looking at the specifications and suggested applications and limitations of those units (e.g. stability with reactive loads) will help you narrow down what you want?

[cat87]

Ok, so, first of all, I think you should provide us with a list of things that you'd like to actually measure, because from what I've read, you're kinda  over-estimating the capabilities of an SMU.

Most of my ambitions when it comes to what sort of circuits I want to use and design relates to circuits well outside my knowledge range, take a SMU for example, although I do know what constitutes a signal chain for collecting samples ending up as voltage or current values on a processor of some sort I really do not know anything about how I would actually use a SMU to analyse a DC-DC converters characteristics for example

If that actually means you want to use an SMU to measure a DC-DC converter, that's kind of overkill. I mean, you could, but an SMU is mostly used for finer things, like testing diode curves, measuring transistor characteristics, basically if there's  a small current to be sunk/sourced, that's where you're using it.

Like other folks said, an SMU is mostly a DC device. If you want something that operates  at kHz or MHz ranges, a 4 quadrant supply is what you want. What that is is a fancy term for a power  amplifier. Something that goes up to about 10 MHz is this. From what I've read, it is  a 100W amp, liner to about 10 MHz

Now, For some resources on SMUs, you could look at the Keithley  236/237 SMU. Schematics are available on the net and it's quite a capable unit.

Try to have a look aver these, maybe they'll help you figure out what you really want to do.