I need to design a high power voltage follower and most of my Googling keeps bringing me here, so although its not an audio circuit, I hope you will forgive me and offer some suggestions.
I need an output range of +/-50V and 50A. A 1 ohm load will be expected so it must be capable of 50A at 50V. The other main requirement is reasonable linearity through 0V. Our first effort using several high powered op-amps just went up in smoke.
I need an output range of +/-50V and 50A. A 1 ohm load will be expected so it must be capable of 50A at 50V. The other main requirement is reasonable linearity through 0V. Our first effort using several high powered op-amps just went up in smoke.
Did you not try some paralled power transistors driven from one of your high powered op amps?
How are you varying the voltage? I.e does it have to swing from - 50 to + 50 at 50amps, or is it dual 50v 50a per rail?
Ian
Hello your project sound like a programable waveform linear power supply, you can do it with a ixys linear power mosfets have a great SOA.
IMO, the first thing you need to do is look at some SOA diagrams for various output devices. Those will give you an idea of what's necessary to do 50A @ 50V. You need a detailed and intimate understanding of what those diagrams mean. Amplifiers are often designed backwards, so from that decide how many output transistors it will take to do the job, then what it will take to drive the outputs, working your way back to the input. Get a copy of Doug Self's amplifier book and Bob Cordell's amplifier book. All will be revealed!
With a really good heatsink and fan you can get about 80 watts of power from a well mounted "150 watt" power transistor. So if you are actually going to want to do 50 v at 50 a you might think you require 35 power transistors on each rail! However if you set your rail voltage to 55 volts then when the output is 50 volts there is only five volts across the output transistors. So the power requirement would drop to 4 transistors. But as you should understand at 0 volts out into a resistor there would be 0 current so no power dissipation would be required. At 27 volts out the power requirement would be 10 transistors but only if the load is purely resistive!
So if you are using an AC waveform half of the transistors would be on at a time so it would be five per rail.
If you want a general purpose power supply that will not fail under any conditions It should really be about 40 power transistors per rail.
One amplifier manufacturer starts out with the minimum number and adds more output devices if they get warranty failures.
For example using ON Semi MJL3281A (NPN) MJL1302A (PNP) A safe minimum gain for the output stage would be 30 so a drive current of 1.67 a would be required. Also use emitter resistors of around .22 ohm on each transistor.
So if you are using an AC waveform half of the transistors would be on at a time so it would be five per rail.
If you want a general purpose power supply that will not fail under any conditions It should really be about 40 power transistors per rail.
One amplifier manufacturer starts out with the minimum number and adds more output devices if they get warranty failures.
For example using ON Semi MJL3281A (NPN) MJL1302A (PNP) A safe minimum gain for the output stage would be 30 so a drive current of 1.67 a would be required. Also use emitter resistors of around .22 ohm on each transistor.
Last edited:
Another option would be a multi tapped mains transformer that keeps the dc input closer to the output thus reducing the dissipation at low output volts at high current.
Just what you need - http://www.diyaudio.com/forums/solid-state/96192-post-your-solid-state-pics-here-32.html#post2332555 Post 1585 -6 ....
Nico Ras has one ("big booty amp" 😀 , next post- above thread) - does at least 40A/4R.
OS
Then it is pretty simple 35 transistors as emitter followers on each rail with .22 ohm emitter resistors. If you want to save money mount them on pieces of black anodized structural aluminum "C" channel around 6" x 24". Use two for each rail, form a tunnel and put a fan at the end. For the driver there are two approaches. The best is to ground the transistors outputs and leave the common (formerly ground) of the bipolar power supply feeding them floating as the output. Then with a second power supply you can use an op amp and a few other parts to drive them.
The more common method is to use either an op amp or differential pair to drive a voltage amplifier stage which drives followers to the output stage.
He said it works , dims the lights everywhere. 😀 Seriously , For 50A/1R- many pairs (15+) of IRFP240/9240 driven by a njw0281-0302 driver pair (huge MOSFET gate capacitance) and a overcompensated blameless voltage stage. With $$ and a day later (and an arc welder trafo) , I would have my 50A AC. PS - is this for a large motor servo ????
OS
Thanks for all the replies, lot to look through. Just a but more info and what I was thinking of trying next. The output will be connected to what is effectively a battery made from several cells, a slow sweep from -50V to +25V is made across the battery. Once I have this working I then need a +/-200V 50A version.
Since I already have the Op-Amps in place I have decided to try the following simple design:
Then parallel up the MOSFETs. The power supplies are variable so I can set them to say +/-51V. For heat dissipation I already have 5 large heat-sinks with fans.
The previous design was 10 OPA541, 1 master driving 9 slaves (as per the data sheet). I had the power supplies set to -54V and +25V, giving 80V range, but the specs on the 541 are different between the web site and the data sheet, so I think I pushed them to far. They where OK at low currents, but blew when I tried 50A.
I've not used MOSFETs before and stuck mainly to op-amps, so when choosing ones for the design what should I look for? I suppose the on resistance needs to be as low as possible. How much so I need to over spec? Would a 55V P channel be OK at 50V or should I look for say an 80V one?
Since I already have the Op-Amps in place I have decided to try the following simple design:
An externally hosted image should be here but it was not working when we last tested it.
Then parallel up the MOSFETs. The power supplies are variable so I can set them to say +/-51V. For heat dissipation I already have 5 large heat-sinks with fans.
The previous design was 10 OPA541, 1 master driving 9 slaves (as per the data sheet). I had the power supplies set to -54V and +25V, giving 80V range, but the specs on the 541 are different between the web site and the data sheet, so I think I pushed them to far. They where OK at low currents, but blew when I tried 50A.
I've not used MOSFETs before and stuck mainly to op-amps, so when choosing ones for the design what should I look for? I suppose the on resistance needs to be as low as possible. How much so I need to over spec? Would a 55V P channel be OK at 50V or should I look for say an 80V one?
on eoutput FET switches to OFF. The other FET switches to ON.
As the polarity of the output changes, either FET can see the full Rail to Rail voltage.
You need at least 75Vds FETS. 100Vds are readily available.
When you advance to the +-200V version, you need Vds>400V
Hi
The circuit in post 15 will have poor linearity about the zero current crossing and probably self destruct because of no gate ESD protection.
So you are making an adjustable battery charger? The description of the circuit function is a bit fuzzy...keep us all guessing... but a linear circuit may not be the best way to go, perhaps a switching circuit?
The circuit in post 15 will have poor linearity about the zero current crossing and probably self destruct because of no gate ESD protection.
So you are making an adjustable battery charger? The description of the circuit function is a bit fuzzy...keep us all guessing... but a linear circuit may not be the best way to go, perhaps a switching circuit?
Yes it works, but was not for me, I am not that crazy. It was designed for a buddy who does not like his neighbors
You do not say whether the output voltage is RMS or Peak, it does make a difference. I have just completed a new product 2.4 kVA sine wave inverter using two very simple bridged quasi-complimentary IRF3710 MOSFETs amplifiers (20 devices), powered by two 12V series connected 560 Ah batteries.
The output current is 100 Amp. The IRF3710 should be connected in parallel (no source resistance) and the electrical path length must be identical both from load and supply so one device does not work harder than another. The amp is not complex at all, but the layout is very tricky at such high currents.
- Status
- Not open for further replies.