EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: JohnnyMalaria on May 28, 2018, 10:43:41 pm
-
Hi,
I need to generate square wave and sine wave signals up to a few 100Hz with amplitudes up to 50V. The maximum current I expect is no more than 50mA.
I have a FeelTech FY2300 DDS function generator that I feed into a unit gain current amplifier which outputs up to 10Vpeak and 3A.
For the square wave, I expect that a bandwidth of 10kHz will allow me to get a good enough square for my application.
Any thoughts or recommendations on how best to approach this?
Thanks.
-
A big chunky audio amplifier should work for the sine.
I am not sure how well it would handle the square.
-
Do you need a constant current output, same as the "unit gain current amplifier"?
Actually, what is that? Is it actually a current amplifier (~0 ohm input, ~inf ohm output, pure current gain within the compliance limits)?
Presumably, the input characteristics won't matter, but for other source and load impedances, you will of course need a different gain, so as to have the same output value. The gain will also not be dimensionless (if voltage input, then transconductance: output amps over input volts). Presumably you're looking for different gain anyway, but do you just need more compliance voltage, or is constant voltage output okay?
Heh, I actually know a guy who designed an amplifier that's adjustable anywhere on the spectrum (CV to resistive to CC), which would probably work here (in terms of required output range), for whatever output impedance you need. Not that that really helps you right here right now. :P Just to say, it's not at all impossible or unreasonable to do.
Tim
-
I guess with unity gain current amp you are referring to unity voltage gain, higher current capability output. Now you need some voltage gain and not much current, or if you don't need DC offset (or you can add it with a different supply) you can solve your problem with a 1:5 transformer at the output of your existing amp, not impossible for a transformer to go beyond 10kHz.
If you get an off the shelf audio power amp you can use it, but for 50Vpeak you are talking (8Ω standard impedance when voltage is the limit) 160W Amplifier, that would be capable of delivering 3A when you only need 50mA. And audio amplifiers aren't DC coupled either.
Getting up to 10kHz 50V/50mA peaks isn't that hard, to start you are well within audio territory. You could start with an audio amplifier design and adapt it for your needs or just from scratch, if you look for 100Vpp you are out of luck with straight IC opamps, but you can brew something with discrete transistors. If you want a simpler option, grab a 200W IC audio amplifier like the STK4050 and build it DC coupled. The advantage is that you don't need a 200W PS, but just enough to supply your 50mA plus the quiescent current (max 120mA for this one) which adds up just to 17W, probably much lower as I'm using maximum rates here and you can tweak the design (bias resistors to taste). So you can build it with 40C+40V 30VA transformer, rectify, bunch of caps and you might get away without regulation.
JS
-
Krohn Hite made some amps that will do it, they appear on the used market frequently enough. IIRC, the 7500 could swing over 100VRMS into 300 Ohms from DC to 1MHz. The newer 7600 and 7602, a little more but they are harder to find. The older DCA-10 can do it 1Hz to 1MHz into 600 Ohms.
-
Do you mean 50 V peak-to-peak max or +/- 50 V?
If it's the former, an output stage formed around an OPA452 would work: www.ti.com/product/OPA452 (http://www.ti.com/product/OPA452)
For the latter, you can use an OPA454.
-
I'd use an audio amplifier for that. 80W (into 8 Ohwm) should be enough to get 50Vpp.
-
Thanks everyone. I should clarify my poor description :)
My application involves applying a voltage between two electrodes (e.g., gold) immersed in a liquid. For the most part, this means aqueous salt solution. At high concentrations, the impedance of the solution can be just a few ohms. At 5V, the function generator simply cannot provide the required current (an amp or so). I use a variant of this:
(https://www.allaboutcircuits.com/uploads/articles/BOPA2_ltspice2.JPG)
where I use +/-15VDC, an AD845 opamp and a TP31/TP32 npn/pnp matched pair. I'm very happy with it.
At the other end of the spectrum, I need to apply voltages up to 50Vpk (100Vpk-pk) in non-polar liquids such as hydrocarbons. Unlike the aqueous case, the currents would unlikely exceed a few 10mA.
I need to apply square or sinusoidal voltages from DC to 500Hz.
My preference would be to find an opamp that can be powered at +/-55VDC or higher.
Since I posted yesterday, I've found this:
http://www.powerampdesign.net/poweropamps/compactmodels.html (http://www.powerampdesign.net/poweropamps/compactmodels.html)
(http://powerampdesign.net/images/200_PAD01_200_pixels.jpg)
which has a built-in fan. It would limit my max. output to +/-40 to +/-45V which is acceptable. The manufacturer also has an evaluation board for it. The cost of both is $126 . Then, of course, I need a power supply such as this from Acopian (https://www.acopian.com/store/productdetail.aspx?q=i1089):
(https://www.acopian.com/images/pic8and25.jpg)
but it's $190 and probably wouldn't provide enough current under my most demanding situations (i.e., high salt concentration water).
I had thought about an audio amp but a complete unit would be far too big and I require DC operation. I looked on eBay for audio amp modules but couldn't find anything with a high enough output.
Thanks.
-
Using kind of a "reverted" architecture, one can build amplifiers that have a very high output voltage range (limited basically by the used output transistors and power dissipation in some resistors), but doesn't need a high voltage OpAmp.
This is an example of a amplifier stage that I've built some years ago, and still use occasionally. Depending on the input resistors (R15, R16), it can output 50Vpk while using a standard amp. It's scalable for higher voltages.
(https://www.eevblog.com/forum/projects/amplify-function-generator-to-50v/?action=dlattach;attach=443728)
X1.1 is the positive supply, X1.4 negative supply, X1.3 and X1.4 supply GND.
X2.1 is output GND, X2.2 output signal. The OpAmp supply (done with a DC/DC converter to isolate) floats +/- 15V around the output signal - this is the trick to achieve the high output voltage though you've got only a standard op amp.
The input signal must be applied from a low impedance source (symmetric or asymmetric) to X4.1 and X4.2.
You' maybe have to re-calculate the resistors R15, R16, R12, R11, R14, R13 for your gain requirements. R12 and R11 take care of the OpAmps input not to exceed its common mode range at the expense of higher noise gain (reduced bandwidh).
Edit: It works virtually with any voltage you apply to the power stage, because of the independent supply for the op amp. So if you need only a 5Vpk amplitude at a few amps, this amplifier does the job with a +6V / -6V supply at vastly reduced power dissipation. No changes required for the circuit to achieve this.
-
This is great - thank so much. The symmetry of the circuit looks very elegant.
A couple of questions:
1. Is there an impedance restriction on the output? If I apply this design to both of my types of load then it will need to range from a few ohms to many 100K.
2. If I choose to use +/-15VDC for both supplies then presumably they can come from the same source? Much of my other circuitry operates at +/-15VDC which I derive from two spare 20V/65W laptop supplies that are connected to buck converters. The outputs are connected in series in the same ways as the two outputs of the DC-DC converter shown in your schematic.
EDIT: regarding GND. My input will be single-ended and the floating GND on my other circuits is connected to "true" GND via the chassis and the other equipment with single-ended inputs that various other instrument cables connect to. So, do R11, R13, R16 and U1B become unnecessary?
-
Note that that amp is class A, or at least pretty deep AB, so it will get quite hot if you need any current from it, and it won't be capable of delivering much more current than it's biased to (leading to clipping and bias shift). This is because the bias is set by the source current sense resistors, and the BJT's Vbe voltage.
That, or it's set by the zener diodes, which then need to be matched to the MOSFET Vgs(th), and mind that low voltage [true] zeners have a really soft curve, so a 3.3V diode at these currents will be more like 2-2.5V, and since bias is through resistors, it will vary with supply voltage.
If you're going to build, I'd just as well suggest an off the shelf power op-amp. Not too expensive, hard to get wrong, and can be configured for current or voltage output. :)
Tim
-
another way (page 7)...
www.analog.com/media/en/technical-documentation/application-notes/an18f.pdf (http://www.analog.com/media/en/technical-documentation/application-notes/an18f.pdf)
-
This is great - thank so much. The symmetry of the circuit looks very elegant.
A couple of questions:
1. Is there an impedance restriction on the output? If I apply this design to both of my types of load then it will need to range from a few ohms to many 100K.
In general, no. You may overload or toast the MOSFETS it if the output impedance is too low (near shorted) and power dissipation get too high. The short circuit current is limited by the small BJTs. It even works into all kinds of reactive loads, I've used it to drive supply rails of DUTs that had more or less large capacitors.
2. If I choose to use +/-15VDC for both supplies then presumably they can come from the same source? Much of my other circuitry operates at +/-15VDC which I derive from two spare 20V/65W laptop supplies that are connected to buck converters. The outputs are connected in series in the same ways as the two outputs of the DC-DC converter shown in your schematic.
Yes, you could use the same supply for the op amp and the output. But this will greatly reduce the maximum output voltage from near to the supply rails using the DC/DC to somewhat like 4V above negative / 4V below positive rail, say 10Vpk for +/-15V rails vs. near 15Vpk for +/-15V rails using the DC/DC. The independent and floating supply gives you the freedom to widely vary the power supply rails to match your experiments requirements.
EDIT: regarding GND. My input will be single-ended and the floating GND on my other circuits is connected to "true" GND via the chassis and the other equipment with single-ended inputs that various other instrument cables connect to. So, do R11, R13, R16 and U1B become unnecessary?
U1B is unnecessary anyway, it's just an leftover since I had that particular dual OpAmp at hand while building the circuit.
The other resistors form the differential input and are still required (at least if you keep using the floating supply DC/DC). Connect one of the mentioned inputs to your GND, the other one to your single ended source. For best results, put a series resistor equal to the source impedance in series with the grounded input. Reversing the inputs is possible, this gives you a simple way to have your signal inverted.
-
Thank you all :)
I greatly appreciate the suggestions - lots to think about!
-
Note that that amp is class A, or at least pretty deep AB, so it will get quite hot if you need any current from it, and it won't be capable of delivering much more current than it's biased to (leading to clipping and bias shift). This is because the bias is set by the source current sense resistors, and the BJT's Vbe voltage.
That, or it's set by the zener diodes, which then need to be matched to the MOSFET Vgs(th), and mind that low voltage [true] zeners have a really soft curve, so a 3.3V diode at these currents will be more like 2-2.5V, and since bias is through resistors, it will vary with supply voltage.
This particular design is AB, with a rather low bias current. Set by the small Zener diodes matching the MOSFETs gate threshold voltage roughly. Interestingly, in this part of the Ugs to Idrain transfer curve one has a lot of allowable deviation from the ideal voltages and still finding a good overall bias. The circuit isn't compensated for temperature variation of Ugs, so the bias varies somewhat with their junction temperature. The small BJTs are for short circuit or "big capacitor applied to output" charging current limiting. In case you use other MOSFET than the ones in the schematic, yes then you may want to adjust the biasing. I've built indeed eight of these circuits for a multi-channel output, and I didn't have problems with bias variations due to component tolerances.
But as simple as it is, it isn't foolproof. Having beefy enough supply rails, one can easily toast the MOSFETs.
If you're going to build, I'd just as well suggest an off the shelf power op-amp. Not too expensive, hard to get wrong, and can be configured for current or voltage output. :)
Tim
Yes, it was recommended within this thread before. And they usually offer self protection. It's a bit hard to find suitable power op-amps that allow for 50Vpk / 100Vpp output, most end up at 80V max. supply.
-
You can bridge an amplifier to get twice the amplitude if the load permits such a setup.