This is what happens when chemists have to do electrical engineering

May 30, 2013 4:40 PM

What's the name of this circuit that I need to build or buy?

I need to be able to switch between two DC voltages (magnitude varied by a pot, but between 0-5 V) based on a TTL signal. I feel like I'm probably missing something terribly obvious in my circuit noobery, but I haven't been able to figure out a simple way to do this? I initially was going to use an AND driver, which would work fine for passing the high voltage (5 V), but wouldn't let me pass a non-zero low voltage (as far as I can tell.) I feel like this is something that has to exist, but I don't even know what it would be called, so googling has had mixed results.

(for reference, this is to be used in a circuit with a 5V DC source, connected through potentiometers set at different resistances to a proportional DC/DC converter. I need to be able to vary the DC voltage going into the converter synchronously with a low-frequency TTL pulse, in order to switch the output voltage. I think another solution might be to feed the 5V into a MOSFET connected to the TTL pulse at the gate pin--basically, using the MOSFET to vary the DC instead of the pots--but I'm not sure I'm interpreting that correctly?)

What I currently have to work with: AND driver, op amp, counter, comparator, XOR gate, converter.

Buying things is an option, especially if there's some sort of IC that will do all of this.

(I am bad at this. Please explain it as you would to a 5 year old, a 5 year old whose only formal education in this thing was the electricity/magnetism section of calc-based physics and who needs a nap after spending two days reading about digital electronics on the internet.)

posted by kagredon to Technology (18 answers total)

You need an SCR or a power FET. (Actually, you need two.) You can't do this with digital circuitry alone.

You didn't say: how much current are your two DC sources going to run?
posted by Chocolate Pickle at 4:54 PM on May 30, 2013

Response by poster: About 300 mA, but that could be halved by having a separate supply for all of the DC/DC converters.
posted by kagredon at 5:09 PM on May 30, 2013

Electrical engineer here. We really need to back up and start at the beginning, because the things you are suggesting, like feeding DC/DC converters from potentiometers, are probably really bad ideas. Unless you are, say, testing/characterizing DC/DC converters, this is not what you want to do.

So tell us, what is the actual application? Something chemistry-related perhaps?
posted by ryanrs at 5:43 PM on May 30, 2013 [2 favorites]

Response by poster: Okay, I feel like I'm probably not explaining something correctly? My understanding is that the DC/DC converters work by generating a voltage that is linear w/r/t a 0-5 V input. I have a 5 V DC power supply, and my understanding is that potentiometers will vary the resistance (and therefore the voltage) on the DC coming out of the power supply and going into the DC/DC converters. (I'm not intending to vary the resistance during use, so if it would be better to just use a static resistor, that's fine.)

I'm applying high voltages (well, relative to 5 V, but it'll be <200 V) to two electrodes in an instrument; I need to be able to switch the voltages synchronously between two different voltages for each electrode, using a 1 Hz TTL signal from a benchtop function generator.
posted by kagredon at 5:57 PM on May 30, 2013

Yeah, you definitely don't want to supply those DC/DC converters from a potentiometer. They will want a low impedance input source, which the pot won't give you.

You say there are two electrodes. Is one fixed at ground, and the other switching between, say 100 and 200V? Or are you doing something like swapping the polarity of the two electrodes?
posted by ryanrs at 6:06 PM on May 30, 2013 [1 favorite]

Response by poster: One's switching -100/+100, the other is +15/+85
posted by kagredon at 6:25 PM on May 30, 2013

Well for the +/- 100V supply, you can just use a DPDT relay to flip the polarity of the DC/DC output. I would feed the DC/DC converter with a benchtop variable power supply, rather than cobble something together from scratch. (I say this as someone who has designed a power supply for commercial production; it's a fair bit of work.)

For the 15/85V supply, I'd use a dual output bench supply and a relay to swap which input is feeding the DC/DC converter.

I take it these two set ups are feeding separate pairs of electrodes in two electrically separated vats of chemical goo? Not one vat with 4 electrodes?

Sorry to pry for details about your experimental setup, but it matters for the electronics. If you could link to a description of the chemical cell you are creating, or in general what is going on here, that might make your requirements more clear.
posted by ryanrs at 6:35 PM on May 30, 2013

Response by poster: They're components in an electron gun in a mass spectrometer, so no chemical goo (in fact, they're under vacuum.)
posted by kagredon at 6:37 PM on May 30, 2013

Best answer: A couple variable bench supplies and some relays to switch the voltages seems like the right way to go. For the +/- 100V supply, there will be on the order of 10ms of dead time as the output relay switches, will this be an issue? For the 15/85V supply, it will take a short time for the output to slew between the two voltages, but the exact time is unknown from the data provided.

Two words of warning regarding EMCO. First, the QH series is unlikely to do what you want (see page 12 of the datasheet for dire looking graphs). Second, the datasheet feels more like a "capabilities brochure" than a list of products available for immediate delivery. You may wish to call EMCO and try ordering a couple devices before you expend much effort designing these DC/DC converters into your experiment. You may find the exact part you want cannot be had in quantities under 1,000 units.
posted by ryanrs at 7:15 PM on May 30, 2013

Response by poster: All right, that works. We've used the EMCO supplies for static voltages before without incidence (they work okay as long as you don't need great precision), but it looks like they're not going to be appropriate for switching. The voltages are low enough that we can using benchtop supplies with the relays should still be pretty cheap (10 ms latency isn't a big deal, since it's going to be switching on a 1 second timescale anyway.)

Thanks!
posted by kagredon at 7:20 PM on May 30, 2013

You can also generate your original signal form using a function generator. It gives you control over the frequency, amplitude and offset of your signal, is variable, triggerable and already wired up.

I've got old Tektronix and Wavetek generators. It would take me two minutes to set up this waveform.

What you do with it after that... entirely different question, but in general, test equipment is the way to go unless you do a lot of EE stuff. This needn't be a design job. It could be a simple instrument set up. Use a T connector and run the waveform into a scope and you can see the output in real time as you use it to stimulate your external stuff.

You won't get 300 mA out of one, or extremely high voltages, but you will easily get your excitation waveform. Look for other test equipment to help with this. Rent from any of the test equipment rental companies. There are a bunch. They have people to help you select the right stuff.

If you are in a lab that is of any consequence and/or at a school with electronics labs, there is test equipment. It isn't all scopes. Look for pulse and function generators.

memail if you wanna.
posted by FauxScot at 7:20 PM on May 30, 2013

I wouldn't say the EMCO supplies are inappropriate, you just need to slap some relays on the output or input to do the voltage switching. But if you have access to variable high voltage supplies (or several hundred dollars budget to buy one), then that's definitely the easier way to go. Cobbling something together yourself is only cheaper if you have EE grad student slave labor or something.
posted by ryanrs at 7:24 PM on May 30, 2013 [1 favorite]

OK, I went back and re-read your post. some of what i said isn't exactly true, but the use of bench test equipment, including function generators, etc. and using rental outfits still holds. custom circuits aren't hard if you've been doing them for a long time, but test equipment has the advantage of being pre-configured for a lot of different functions. if you can turn design from a tedious fab effort into a simpler configuration problem, it's often faster and better.
posted by FauxScot at 7:26 PM on May 30, 2013 [1 favorite]

Best answer: With the 1Hz switching and the fairly high voltages, I'd guess you're not going to have much joy with mechanical relays. Or at least, not for long. There are solid-state relays, but I'm not sure if they'll work as you intend when (I suspect) you have some unusual (i.e. non-resistive) load currents.

Can you give the part number or schematic or datasheet for the electron gun? If you're changing the relative potentials of two plates or grids or something inside there, then it seems like you won't have much current flowing. There'll be a little spike each time you switch, as the plates fill up or empty of electrons.

Hmmm, looking at a couple of messages ago, you're switching one electrode +/-100, and the other between 15 & 85. You can do this with a +/- 100 dual supply, and divide the 15 and 85 off the +100. I don't know a cheap & easy way to do this offhand, but maybe ryanrs can deliver. Or just use a bunch of supplies.

How about this: set up a 1Hz square-wave generator using a 555 or something. Use optoisolated FETs or transistors to switch the high and low voltages. If you want to get fancy, use a microcontroller to generate the square waves, so you can avoid overlap between the high and low-side switches.

Here's an optoisolator that might work.
posted by spacewrench at 7:28 PM on May 30, 2013

Any relay rated for 120V AC (which is probably most relays) will be fine. 1 Hz switching is likewise fine.
posted by ryanrs at 7:39 PM on May 30, 2013

Response by poster: The electron gun is...kind of slapped together from a few different spare parts, but the actual electrodes in question are to gate the electrons and the ions that are generated by them (and shouldn't have much current; the 300 mA figure I cited earlier was a worst case for maximum load conditions on the EMCO converters.) I am already using a function generator (an SRS model, I think?) and benchtop pulse generators, but the pulse generators are meant to be for high-frequency, high-power applications, and it's not too practical to be tying two up to do 1 Hz switching at relatively low voltages (compared to what they could run.)

The optoisolator idea seems like it could work, too, and I think I know someone I could cadge a few off of if the benchtop DC supply solution doesn't pan out.
posted by kagredon at 7:41 PM on May 30, 2013

Best answer: It kind of sounds like you have two possible approaches— you can use two high-voltage supplies and switch between them, or you can use one "programmable" supply and vary its voltage using its 0-5v control input, right?

Option 1 uses more hardware and is definitely inelegant but might be the easiest approach. A 300mA / 100VDC mechanical relay is not hard to come by. (Note the VDC; relays tend to have much higher AC ratings than DC.)

As for option 2, I would expect a decent function generator to be able to generate a square wave that goes between two arbitrary voltages for the control input, and if you can arrange that, it's probably the best way to go for an experimental setup. Otherwise, a couple of approaches come to mind:

• Switch between two low-voltage signals using an analog switch (eg the venerable CD4053 or CD4066, or maybe something from this century if you prefer);
• switch one leg of a voltage divider in and out of the circuit (followed by a buffer if the power supply's input needs a low impedance) — this circuit is the simplest I can think of, but probably isn't accurate to more than 5-10%.

posted by hattifattener at 12:59 AM on May 31, 2013

Best answer: I'll try to describe this with words, but feel free to send an email if you would like a schematic:

* Put a CMOS buffer on your TTL signal. That could be your AND circuit with shorted inputs, but it's important that it's CMOS rather than TTL, otherwise its output voltage is insufficiently well-defined. The buffer is powered by your +5V source.
* Put your potentiometer between the buffer output and your +5V supply. The pot can be any value, just high enough that there won't flow significant current through it. 10-100k should do.
* The wiper of the pot will have a potential of +5V if the TTL input is high, an the low voltage is tuned by the pot position. If I understood correctly, this is what you want.
* This is a high-impedance potential, and you need to buffer it. Search for op-amps that work rail-to-rail with a single supply, are stable at unity gain and can supply high output currents. Connect the negative input of the amplifier to its output and the positive input to the pot wiper. That's a unity-gain buffer, and the output of it goes on to your DC-DC converter. Just googling quickly, the AD8531 can source 250mA and meets the other specs.
* Since you drive an electrostatic gate, the impedance seen by your DC-DC-converter is purely capacitive. If you have some sense of how big that capacitance (Cgate) is, you can put a resistor (Rgate) between the high-voltage output and the gate. That will decrease the inrush current during switching. You want that resistance low enough that Rgate*Cgate is much lower than the required switching time.

If your components are in vacuum, please take into account that they won''t cool very well.
posted by springload at 8:15 AM on June 1, 2013