Hello allaboutcircuits!

Hope somone can point me in the right direction here..

I have to make a digital circuit that via a 16 bit DAC can control a powersupply that goes from 0-1000 VDC. I have made the digital circuits and the DAC is "outputting voltages" that goes from 0-5V. I have some thyristors (BT137-800) that is my intend to put in series so they can handle 1000V. I don't have much experience with thyristors, so now for the problem:

"How do I get the 0-5V to be represented as 0-1000V with the thyristors?". I can imagine I somehow have to use them as amplifiers... Is that possible?

All the best

Ole.

 

Some information about the power supply is necessary - especially how the output voltage is controlled. Is there some control for this purpose?

No thyristor is useful for level control of a DC voltage. The action is either on or off. Once triggered, a thyristor will remain in conduction until current through it has fallen below some minimum level. Thyristors are AC control devices.

 

Hello beenthere.

Thanks for the reply.

The powersupply has to be controlled explicitly via the digital circuit/DAC, so it has no control.

(I don't have the powersupply yet (I not fumbling around with 1000V, but the circuit have to be able to handle it, so right now I'm using a 30V suppply)).

 

Hmm..guess I'm will have to get some power transistors of some kind instead?

 

It's too bad, but "the power supply" you mention is not meaningful. Does this power supply have an actual existence, or are you being asked to design one that may be controlled by the DAC output? There are such things as voltage controlled power supplies.

The nature of this assignment is not clear.

 

I have to design the controlling part of the powersupply. The voltage from 0-1000V is to be controlled by the circuit I'm building. So I connect a 1000V powersupply externally and the digital circuit with the DAC has to control the/a voltage from 0-1000V. (hope it is more clear now)..

 

What is going to be your approach to the type of power supply? Or does that mean you have to swap your controlling circuit for the one already in place in an existing supply? Sorry, it's not clear yet.

Will this be an actual device, or is this a paper exercise?

 

I'ts no paper exercise. There will be an actual device. I will have to build it so the power supply connects externally. Right now when I'm experimenting and doing the basic calibrations, I will use a powersupply that goes to 30V, but later I will exchange that with the high voltage one.

What is going to be your approach to the type of power supply?

As I'm not going to build the supply itself I don't really bother what kind it is. It's probably in the first place going to be a high voltage lab-powersupply that is set to 1000V.

But I think you already might have answered my question. I had some thyristors and assumed they could be used for amplification but I can see now that they are rather used as relays. So I guess I have to get eg. some power-MOSFET's (if it's possible to get some that can handle those voltages) .

 

We have done something similar at a lab course at my class. As the thyristors are only capable of inducting current only at positive voltages (like a diode) and only when a positive current signal is given at their gate, they are quite useless in a DC situation. What we had seen at the lab, was a rectifier circuit, (a bridge type I think), wich was powered by the mains. What the thyristors had to do was to take the place of the diodes in the bridge, but allow the sine wave to come through them only past a certain angle (between 0 and 180). This way, only a portion of sine was coming out of the bridge, and thus the effective work was reduced as desired. This was reflected as a user-selectable DC voltage coming out of the rectifier. This however demanded a complex control center for the thyristors, as it had to keep them in synch with the mains frequency and operate them precisely at the desirable angle in the period of one cycle.
So, my conclusion is that, if you are given an AC voltage, and you are asked to produce DC, you can replicate my lab equipment, otherwise, I don't see how a thyristor can help you.
If you are given a DC power supply, you could turn it into a pulse voltage. 100% duty cycle for 1000 volts, 0% duty cycle for 0 volts. But this kind of supply would be dangerous for anything else other than purely resistive loads.

 

Thanks Georacer! But I don't think I need that. I have ordered some power MOSFETs that can stand 800V, my plan is to connect them in series so they can handle the 1000V and then use them to amplify the 0-5V signal into 0-1000V. Guess that should be possible somehow..

 

You might want to study up on those FET's. The gate to source voltage is the device control, and it has to remain within clearly stated limits.

You might want to look into IGBT's.

Have you got a design to post?

 

You might want to study up on those FET's. The gate to source voltage is the device control, and it has to remain within clearly stated limits.

You might want to look into IGBT's.

Have you got a design to post?

Nope, but I got a Sedra and Smith book, and some old courses I'm going to dig up. I have worked with MOSFET's circuits before, a while ago..

If I get stuck I might return

 

By the way I imagine the structure of the circuit something like this (no values on it yet)..

 

There is only just enough voltage on the gates to place the FET's into threshold conduction. Think also what happens to M2's Vgs when any current makes a drop over R2. Or, for that matter, look at Vgs with M1 when it and M2 (Q is more a conventional designator than M, by the way) conduct. To control the FET's like that is impossible.

Replace the FET's with a short, and look at the voltage across Rl. It's not zero. With the FET's off, the voltage out is the voltage between R1 & Rl. Your shunt does not do very well, either.

 

Ok, have made the model simpler (The 10Vdc is actually a summing amp circuit, able to offset the DAC with VT, but let that be, I am interested in the priciple). Right now I am running with V1 = 30VDC (I know I would have to re-measure/calculate later with a higher voltage).

I have measured ID-VGS for different values, calculated k and VT from (not sure what it's called in english) the 'two equations with two unknown quanties' - directly translated.

VT = 3.47V which seems reasonable from measuring, and from the datasheet that says VT = 3-5V.

Now, normally when MOSFET's is used as signal amplifiers, I guess one would find the Q-point, for the biasing, but I am to use it as a "DC amp". I have read that the transconductance gm is ID/VGSQ, But I would have to use the MOSFETs rather from VT to "Max" instead of having a bias holding it at the Q-point.

Before I have gm, I don't know how to calculate the gain accurate.

I have tried experimenting with the values of R1 and RL, but it seems that no matter what, I can only control the MOSFETs with in a range of approxsimately 1V (which isn't a problem if it is so, cause I can just get the DAC to work in a smaller range. I just wonder if it has to be like that?).

 

Is that 12 Meg resistor the actual load for the controller? In that shunt setup, the 20K resistor would be dissipating 50 watts when the FET's were fully on. Is this acceptable?

 

Is that 12 Meg resistor the actual load for the controller? In that shunt setup, the 20K resistor would be dissipating 50 watts when the FET's were fully on. Is this acceptable?

Well right now the 12Meg is the actual load (but easily exchangeable as the setup is on a breadboard, it's just a starting point).

The circuit have to be able to make "curve traces" on transistors and vacuum tubes (which for the last part explains the need of the high voltages). I'm not sure how much current they are going to draw yet, for now I'm just trying to understand what's going on, and how to calculate the values..

 

Okay, a curve tracer. You probably will not need 1 KV applied to anything, unless you plan on duplicating data sheets for 1200 volt IGBT's. I guess you might need that voltage for dynode strings on PMT's.

I would be nervous about the fixed 1000 volts, as any failure or glitch in the shunt means that the full voltage is present on the test fixture. If you are planning on doing curves on big FET's, you will also need lots of current capability (power transistors too).

Were I to do this, I might want to feed the power supply (or perhaps just a rectifier and filter cap) witha controlled voltage. If you have the step up transformer, using a variac to feed it lets you set the output voltage over quite a range. Then you only need to control the base current/gate voltage/grid voltage to obtain the output curve. Your 16 bit DAC might do better at that function.

I dug up references to some transistor and tube curve tracers:

http://www.repairfaq.org/sam/semitest.htm

http://www.vintage-radio.info/heathkit/

http://www.tech-diy.com/MCTracer_no_pix_files/MCTracer_opto_ver_feb6_04.htm

 

Okay, a curve tracer. You probably will not need 1 KV applied to anything, unless you plan on duplicating data sheets for 1200 volt IGBT's. I guess you might need that voltage for dynode strings on PMT's.

I need the 1KV voltage for big vacuum tubes, or to be more precise the professor who set up the assignment needs it..

I would be nervous about the fixed 1000 volts, as any failure or glitch in the shunt means that the full voltage is present on the test fixture. If you are planning on doing curves on big FET's, you will also need lots of current capability (power transistors too).

Yes, I have thought about that; if the controlling circuit is off, the powersupply is at max voltage, due to the inverting circuit. Maybe I'm going to put in a relay, to fix that.

Were I to do this, I might want to feed the power supply (or perhaps just a rectifier and filter cap) witha controlled voltage. If you have the step up transformer, using a variac to feed it lets you set the output voltage over quite a range. Then you only need to control the base current/gate voltage/grid voltage to obtain the output curve. Your 16 bit DAC might do better at that function.

It has to be in a way so the stationary psu is connected externally, so I can't let the system control the psu directly (I'm not quote sure if that was what you ment?). I'm not going to work directly with the 1KV, I will make the system to work with 60V, re-calculate the resistances and make some controlled tests with the high voltage supply later.

Thanks for the links and answers! I will see if I can extract the information I need..

 

By the way the MOSFETs I'm using is two FQP3N80C powerMOSFETs (the resistors on the drawing will represent several resistors in series to handle the Watts).