Hello guys. I've built a voltage controlled current sink. I'm using the basic op amp - transistor configuration, and so far it appears that it is working as expected. I'm applying a square wave at the op amp input to obtain current pulses.
The problem is I need to be able to invert the direction of the current at the load. For this reason, the first thing that came to mind is an H-Bridge. I've found a paper that used such configuration. However, the voltage/current levels are different.

The load is a 1k resistor with a small parallel capacitor (500pf). I need to provide 100mA pulses (max), so the load will see 100V max.
My questions are: Is this configuration a good approach to deliver the pulses in both directions? Are there any H bridge IC's that might fulfill such current/voltage requirements? If I need to build it myself, what other points do I need to consider for this approach.
Thanks a lot guys, any help is appreciated.

 

Another approach is to use something like a Howland Current Pump (shown below). In that case, you would use a high current op amp and to invert the current, just invert the polarity of the pulse with respect to "ground". The advantage is that the polarity switching would be done in a low power stage.

http://www.indiabix.com/electronics-circuits/howland-current-source/

Or, if you prefer YouTube

 

I think it will work ok, but you will need to protect from the gate source voltage being to high with some zener diodes and resistance in the gate.

 

Sorry, I forgot to say that the Load must be floating, so the Howland Current Source doesn't quite do the job. Also, I once looked but count find any op amps that can handle those voltage levels.

For rony: which gates are you referring for the zener diodes protection? Also, you mean I should add a resistance from the digital pins to the gates, or do you mean a resistance between Q4-Q5, Q6-Q-7. How do I go about calculating their values?


Thanks a lot!

 

Not as simple as it sounds - but maybe I'm making it hard.
Do you want to be able to adjust the current?
Is the op amp how you want to control it or do you want to control it with the FETs?
Anyway have a look at this one.
I moved the current source to the low side because if it was in the high side like you had it the op amp would need to handle the high voltage.
Made it adjustable - the input to the op amp is 2 ma per volt.
The FETs control the direction of flow.

 

Just a quick note. The arrangement of U1A and Q1 are such that the analog voltage to the bridge cannot get up to 20 volts, let alone make use of the +100V on the collector of Q1 because Q1 is a voltage follower.

You may want to revisit the analog output stage before laying this out or building the circuit.

 

OK guys, thanks a lot for the information, I think I'll try something like ronv suggested. Thanks for the help, as soon as I have an update on anything I'll come and tell you guys!
The op amp is how I want to control the current by the way.
Thanks a lot.

 

Ok, after analyzing the circuit I have a couple of questions: The BZX84C12L zener diodes are there to make sure that:
1. VGS for the IRF9640 (P-channel) is -12V max, and VGS for the IRFR120 (N-channel) is 12V max correct?
2. What about the resistors, I mean, why use 10, 20k resistors, Im asking because the switching times will be about 300µs for the fastest required, and I know that the Input capacitance has to be taken into account, but I've never done these calculations before. How can I go about calculating the switching times with these values?

Thanks a lot guys, appreciate the help!

 

Oh and also, which software do you use for simulations? thanks!

 

Ok, after analyzing the circuit I have a couple of questions: The BZX84C12L zener diodes are there to make sure that:
1. VGS for the IRF9640 (P-channel) is -12V max, and VGS for the IRFR120 (N-channel) is 12V max correct?

That's correct.

2. What about the resistors, I mean, why use 10, 20k resistors, Im asking because the switching times will be about 300µs for the fastest required, and I know that the Input capacitance has to be taken into account, but I've never done these calculations before. How can I go about calculating the switching times with these values?

That was why I ask who controlled the current. If the op amp controls it I thought the direction not to be to important. But having said that you can get close by looking at the data sheet got total gate charge (nC) this is the time it will take to switch with 1 amp of gate current to discharge the gate capacitance. Since they are low current Fets they have low gate charge - I think 6.7 NC. so 6.7 X 10-9 / .0006 (1/2 of 1.2ma.) amps = ~ 10 usec. They never switch that fast but it gets you close.
While I was looking at this I noticed I didn't add any gate to source resistor to the lower FETs to turn them off so please add one. Also noticed the clamp current goes thru the current sense resistor. This is an error term. So making the gate drive resistors even larger for the lower FETs would reduce it.
If you are going to use the H to switch the current there is a problem with the response time of the op amp. This makes the current high (er) until the op amp responds. If you control the op amp it is smooth but the response time is about 25 usec with the 358. A faster op amp would improve on this if you need it.
The program I use is LtSpice. It is a free from linear technology and very popular. It is really handy because you can see the nominal rise and fall times, etc.


Thanks a lot guys, appreciate the help![/QUOTE]

 

Ooops, I see you are using different FETs, so different gate charge, but also not good for 125 volts. You need some extra voltage for the sense resistor. Let me look for some good ones.

http://www.mouser.com/ds/2/200/irfr220npbf-223238.pdf
http://www.mouser.com/ds/2/149/FQU5P20-195483.pdf

 

If you want the load has 100V max to output, the Q1 need to change to as an npn bjt to drive the pnp power bjt.

 

But having said that you can get close by looking at the data sheet got total gate charge (nC) this is the time it will take to switch with 1 amp of gate current to discharge the gate capacitance. Since they are low current Fets they have low gate charge - I think 6.7 NC. so 6.7 X 10-9 / .0006 (1/2 of 1.2ma.) amps = ~ 10 usec. They never switch that fast but it gets you close.

I just have a question here, where does the current 0.0006 mA, or the (1.2mA) value that you use for calculating the time comes from? If I'm getting things right, I think I can connect a digital pin that outputs (5V, 20mA(max) ) to control the mosfets. I could add a resistor to limit the current to 10mA, and using a fet that has 21nC of Gate-Source Charge, Im looking at 21nC/10mA = 2.1µs to switch?

I've installed LTspice and right now Im trying to get the same simulation going, I'll update with some info later.

Thank you guys.

 

Since the gate to source voltage is 12 volts (zener voltage) and the resistor is 10k, 1.2 ma.
Here is the spice file, maybe save some work. Don't think you can drive them directly since the source is not at ground.

 

Since the gate to source voltage is 12 volts (zener voltage) and the resistor is 10k, 1.2 ma.

Well now that seems obvious, thanks !

Here is the spice file, maybe save some work. Don't think you can drive them directly since the source is not at ground.

Ok so I got the simulation going. LTspice seems really useful,(not necessarily practical hehe). Why wouldn't I be able to drive the Mosfets with a digital pin? If I measure the current at the mosfet gate without any resistor in them, it comes up to +-2Amps aprox. If I limit the current with a resistor to a value that the digital pin can handle (R = 330 -> I=15mA), and the circuit appears to work as expected. I don't see what the problem will be with that approach. What do you mean that the source is not at ground? It can be controlled directly with the digital Pin value.

Will I need to get a gate driver? I think I could get my hands on the TC4423A.


Thanks for the file too!

 

Not sure we are talking about the same thing. If we use the bottom right FET......
For a current of 50 ma. the left side if the 1k will be at 125 volts and the right side at 75 volts. From this the drain and source of the FET needs to be at 75 volts. A 5 volt signal can not turn it on.
I don't think you need a gate driver, but we need to understand how you want the circuit should work. If you want to reverse the current thru the 1k at some rate - say 5khz and just set the current level with the op amp we will need a different approach. I went off your first schematic that showed the current pulses on the op amp, but I bet that was a bad assumption.
What is the load?

 

Well, I need to alternate the direction of the current pulses through the Rload. The opamp recieves its +Vin from a DAC circuit, and I 'm planing on using some digital pins of the arduino to control the H bridge. The pulse width of each pulse needs to be from 250us to 500us, and the frequency goes from 20 to 100Hz max.



Would using a higher voltage, say 9V-12V do the job of turning it on?

 

Ok. Here is a picture of the gate and source voltage.
Also the current overshoot in your load.
It works ok the way it is because the top left FET turns on the bottom right FET and vise versa. If you wanted independent control you could add a transistor with the collector going thru a resistor to the 125 volts.
I attached a new file with an op amp from the library since I just realized you probably don't have an lm358 model.
I don't know if the current overshoot is a problem for you.

 

OK, thanks a lot man, I think I'll do some tests. You where right I didn't have the lm358 library, but I've managed to download and use one. I don't think the overshoot will represent a mayor problem since 100mA will be the highest required value, but Im not quite sure of this yet. Thanks a lot man, this is a great starting point.

 

One last thing.
You will need a bigger transistor at the bottom. Something like this:
http://www.mouser.com/ds/2/149/KSC2690A-88885.pdf