Amplifier Schematic

Hello All,

Could anyone here direct me to a schematic for a high current, low voltage signal amplifier? I need an input Voltage of no greater than

8V, and an output current of about 20A. Sone Voltage gain is acceptable. The wave to amplify is a square wave. I know this is a tall order, but I have an existing H-bridge circuit that is acting as a square wave amplifier now, and I don't want to use it for too long, as the signal is attenuated at low Ohm loads. I have at my disposal several high current rated MOSFETs, and all types of standard resistors, inductors, capacitors, etc. Thanks for the help.

I'm thinking if you want a square wave at 20A, you need to be very specific about the slew rate.

I am not going to run this thing at a frequency of any higher than 100 Hz, possibly 50 Hz. The slew rate as I understand it in this case would be analogous to the switching time between the two signs of the wave. This square wave doesn't need any voltage offset. I am new to the concept of slew rate, as I have been trained in ME, not EE unfortunately. If what I am getting at here is correct then a slew rate of 0.04 V/us would be acceptable, which would be about 10% time of a

100 Hz wave. But then again I could have this concept totally wrong.

Do you need it to be linear? Do you need the output current to be either 0 A or 20 A, or all possible values in between?

If the input is always a square wave, best is to use a switching amplifier, as you were using. This is why I asked about linearity. You don't seem to need linearity.

I guess you mean that, at low ohmic loads, you don't have enough load current. That might be due to:

- Insufficient supply voltage for a given load resistance.

- Excessive load resistance for a given supply voltage.

- Excessive switch resistance (make sure you use low resistance MOSFETs, and that Vgs is high enough).

1, 2 or 4 switches (MOSFETs / BJTs, ..) is your best structure, if the input is always a square wave. A linear amplifier won't help.

As far as the Amps are concerned, I am driving a 1/2 Ohm resistor essentially with 0 to 8 V. So I guess you could say I need all the Amps between 0 and 16, I put 20 for a margin of safety. I have a DC power supply that can source about 100 Amps. The H-bridge I have works fine at 4 Ohms and 8 V, but attenuates voltage and current at 0.5 Ohms and 8 V. The gates of the mosfets are activated by 4V or so, and I have the supply powering the gates, so I guess I can't get under like 5V out of the existing set up. This should be ok for the appication though. Its a

4 MOSFET switching unit, and the transistor current rating is 110 A drain-source. I think that the Rdson is too high or something in the MOSFETs at this low ohm level. I am not sure why this is happening, but it seems to be associated with the resistance of the load. I figured maybe an amplifier of some sort would be a better choice than the h-bridge due to this phenomenon. I'm going to mess with the h-bridge this morning, hopefully I don't burn any of these MOSFETs ...

Yes, that sounds like Rds is too high.

Can you tell us which MOSFETs are you using?

Best,

But only at the edges of your square waveform. Never in steady state, right?

A two-level square wave that represents a real world physical magnitude has of course infinite levels, because real-world physical magnitudes cannot be discontinuous, and so at the edges of the square wave of i_load(t) you will have all the intermediate levels.

In your case, you will have, at some time points, i_load(t)= 10 A, but that does not mean that your amplifier has to be designed to be able to provide i_load=10 A **in steady state**. You do not need that [if it is true that the input is always a square wave]. And not needing that, relaxes the requirements for your amplifier. It does not need to be linear. It is best if you use a switching amplifier.

Best,

The MOSFETS that I am using for the h-bridge are IRF4905, IRF3205. Spec sheets for either of these can be found

and respectively. I hope this sheds some light on the subject.

You need higher Vgs values, with the MOSFETs you are using.

Look, for instance at IRF3205.page3.figure3. Vgs=4 V is even out of the picture.

Or IRF4905.figure3. With Vgs=-4 V and T=25 ºC you would only get I_d=-3.5 A @ Vds=-25 V and about I_d=-7 A @ Vds=-50 V. You need I_d=-16 A @ Vds=-48 V. So you are not even half way there.

Also, look at figures IRF3205.page4.figure6 and IRF4905.figure6. You need to choose a |Vgs| value above the horizontal (or almost horizontal) line in the middle of those two figures. That is to guarantee that you fully create the channel. Minimum would be around 6 V, but I would use 10 V or so, to fully take advantage of the low resistance capabilities of your MOSFETs.

Best,

Okay, I think you need to be a bit more clear about your application. Based on the information you've given thus far, the problem is either very hard or relatively easy; the issue is that we don't know what matters and what doesn't.

Here are some of the things that are unclear:

1. You say you want a "square wave". Now, a square wave in principle is something with infinitely fast rise and fall times, but in the real world there's some limit; that's the "slew rate". Switching large currents quickly is challenging, because of inductance. But, switching large currents very slowly is also challenging, because of heat dissipation. So, if you just want the output to switch fully on and off as fast as it conveniently can - that is, if the slope of the turn-on and turn-off don't matter - that's one thing; if you need a precisely controlled rise and fall time, or if you need it to precisely track the input waveform, that's another. Which is it? And if it's the former, exactly what are the rise and fall time requirements?
2. You say you want an "amplifier". Usually we take that to mean you want something whose output voltage is a linear function of its input voltage. But then you say you want a square-wave, and you're using an H-bridge; that makes it sound like you're doing PWM (pulse width modulation), in which case you don't care about the linearity of the relationship of output to input, you just want the output to switch ON when the input is ON and then OFF when the input is OFF. Which is it? Switching is a lot easier than linear amplification.
3. What kind of load do you have? It makes a difference whether it's inductive (like a motor) or resistive (like a heater).

I'm going to GUESS that you don't need an amplifier at all, you just need a switch; that your output voltage can be controlled by adjusting the supply voltage, rather than by adjusting the input voltage; and that your present problems are that your MOSFETs don't have low enough Rds and also that you are using the same power supply to drive the gate as to drive the load. On that latter point: you'll find that if you use a separate power supply, then you won't have the problem of the gate voltage being too low for reliable fully-on switching.

But that's just a total guess. Please be more clear about what's going on.

OK. I have an 0 to 8V power supply that can source ~100A. I also have a function generator that can provide signals from 0 to 10V, which I have set to a square wave from 0 to 1V. Both of these components are powering and controlling an H-bridge, respectively, that I made. The schematic that I used for the bridge can be found

It has been slightly modified at "B" to accept the f.g. input. Where the schematic says +12V and ground is where the power supply is being connected. The function generator shares the same ground. The modification was as follows

| < 10k Ohm |

to B - -|

C 47k Ohm

NPN B---^^^---- f.g. in

E |

GND Also D1, D2, D3, D4, R7 have been removed. The power supply is being used to provide gate voltage to the H-bridge, and the Voltage across the load. The f.g. is being used to activate a switching transistor setup to control the gate voltage. The load that this thing is driving is about 0.5 Ohms, with negligible inductance. I am sure the load has very small inductance. It is important that the square wave does not vary on a set frequency. The frequency range of the circuit is from 0 to 100 Hz. The slew rate can have a good deal of leeway, I'd say the rise and fall time can be as long as 10% of a period, although the faster the rise time the better. The load voltage should be as close to the power supply voltage as possible. This voltage is being attenuated at high Ohm loads, it works alright at 4 Ohm test loads but not 0.5 Ohm. I originally started this thread looking for an amplifier, because I thought that it may have been a viable option, but it looks like a switching circuit is the best way to do this. I think that I may be able to modify the circuit to accept another power supply for gate voltage, but I have no other power supply that can do high current other than the one I have which is limited to 8 V. I've been told this h-bridge circuit isn't particularly well designed, although I'm not sure why. I guess that the best thing to do here is to modify the circuit to accept two power supplies. If anybody knows of a better h-bridge design, please let me know. Thanks for the time!

Sorry, I have been mixing up numbers from two threads. :-)

Of course your Vds is not 48 V. It is 8 V. Anyway, that makes it even more difficult to get the load current you want, with the Vgs that you provide. You need higher Vgs values, to cause lower Rds values. |Vgs|=10 V, as I said, would be good. If your supply is 8 V max, and you can only generate Vgs from it, that would still be ok. However, the MOSFET drivers I see in your pdf are not very good. You charge the input capacitance (to the MOSFET) actively, with a BJT, but you discharge it passively, with a high-value resistor. The turn-off times will be too long, and especially knowing that your MOSFETs have a high total gate charge Qg requirement. I don't know your switching frequency, but I recommend totem-pole MOSFET drivers.

Best,

Yes, sounds like switching is what you want. If I may restate your problem:

You need an H-bridge driver that can drive the MOSFETs fully so that their Rds(on) is much less than the 0.5 ohm load resistance, at currents up to

20A. The H-bridge is to be controlled by a square wave at 50 to 100Hz. However, the high-current load supply voltage varies from 0 to 8V.

Here's another question: how *low* a load voltage do you need this to work with? Do you want to be able to go arbitrarily close to 0V, or are you just interested in the range around 8V?

If the lowest voltage you need is on the order of 4V, then you might be able to find ("logic level") FETs that will turn on at that low a gate voltage. But if you need to get all the way to 0V, then your bridge driver will need a bipolar supply; it will need to be able to drive the gates of N-channel FETs more negative than ground. (Basically, if "Vp" is the voltage of the power supply powering the load, and "Vgs(on)" is the gate voltage required to turn the MOSFET fully on, then the bridge driver needs to be able to take the NMOS gates to Vp - Vgs(on). If Vp is, say, 1V, and Vgs(on) is 10V, then you need to get to -9V.

The thing that makes H-bridge controllers tricky is that if both the FETs on one side turn on at the same time, then you've got a short to ground and you get to briefly test whether your supply can really put out 100A. (P = E^2/R; if the FETs have Rds(on) = .01R, and you put 4V across 'em, you get

1.6kW. That's when you discover the difference between convection and invection.) That's why the other fellow was concerned about the difference between turn-on and turn-off time. The current circuit turns on the top FET faster than it turns off the bottom one, so there's a moment there where they're both turned on.

If it were me, for that reason I'd consider using an IC H-bridge controller rather than doing it with discrete components. The integrated ones have special protection circuitry to ensure that "shoot-through" doesn't happen.

Anywhere between 5V and 8V should be find for the supply voltage. Do you know of any IC's that would control the MOSFETS using an input Voltage pulse from the f.g. that will control the switching frequency? I am going to look into that.

Google for "h-bridge driver" and you'll find a zillion. Look at TI, IRF, National, Zetex, just to name a few.

You might try posting a question on sci.electronics.components, asking for a recommendation for an h-bridge driver IC and compatible MOSFETs that will work in the 5-8V, 20A range, with switching frequency 50-100Hz.