Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Ninety six separate , isolated, high-side NFET gate drive voltages required

Status
Not open for further replies.
Please assist…We have 96 back-to-back NFET switches which are to be switched ON/OFF occasionally. We are using 30V NFETs (3x3mm). They will carry 10 Amps when ON.
We need 96 separate High side isolated drive voltages (~12V) to switch all of these.
What is the cheapest/smallest way to get all these high side isolated drive voltages?

Transformer drive chips..
We could use 96 Max845’s with 96 Halo TGM-010P3RL transformers with voltage doubler rectifier on each output. However, I reckon the TGM-010P3RL will likely go nil-stocked…..so I am also pondering winding ninety-six 1:1:2 transformers myself on torroids.

Flyback converters:
However, I am wondering if I could do eight flyback converters, each with 13 separate isolated dsecondaries (one secondary for primary side regulation). I could make it a “triple layer sandwich wind”, with a primary of 4 series layers, with three secondaries in each “primary sandwich”.

What do you think is the “lowest component count” way of doing this?

MAX845 transformer driver

TGM-010P3RL transformer
 
Thanks, those look really good, though the Vout might only be 5V in some cases. We would like a good 10V to drive the fet to ensure lowest rdson.
 
What is the supply voltage that you will get the 10A from. 10Ax96=960A
MAX845 is a option. Do you have 5V to 3V to run it? With the right transformer you could make a +12V supply that sits on top of you main supply. If you don't have 5V for the MAX845 I know of a part that will do 30 volts. LT3999.

Q1, R3, R4 = one part. "Digital transistor"
1601336180283.png
 
Thanks, those look really good, though the Vout might only be 5V in some cases. We would like a good 10V to drive the fet to ensure lowest rdson.

inexpensive nFET exist with R(on) less than 5mOhm when driven with logic level voltages (3 to 5v).
 
Thanks for the LT3999, that looks like it could be used in the attached, which is a way to produce 4 isolated outputs...possibly more.
Please tell what you think is the maximum number of secondaries that might be possible here?

BTW, now they are saying they may want to PWM the FETs sometimes at say 1kHz.
All FETs must be individually controlled.
 

Attachments

  • Push pull transformer driver.asc
    11.1 KB · Views: 165
  • Push pull transformer driver (1).pdf
    34.5 KB · Views: 188
I like the circuit ronsimpson showed in post #4. That will give you independent control of each channel with only a single isolated power supply. And, if the input voltage is stable, you can use a boost converter to make the high side gate voltage instead of an isolated supply.

Be aware:
1) The enable signal is inverted. A high in R4 will turn the channel OFF. A low on R4 will turn the channel ON. This means that, assuming that you don't want all 96 channels to be ON at system power up, you need to ensure that all Q1s are ON at power up. This can be done by connecting R3 to +5 instead of GND.
2) The turn OFF time will be fast. The turn ON time will be slower as the mosfet gate gets charged through R2.

Edit. Corrected a mistake in this post, See post 11 below.
 
Last edited:
Back in post #9, I said that the gate drive voltage could be generated by a boost converter. That is incorrect. The gate drive voltage does need to be floating and referenced to the positive power rail as shown in post 4.
 
What's the actual configuration of the FETs? In the description, I see "back to back" and "96 individual isolated drive [supply] voltages", which seem conflicting.

If they're back-to-back, and therefore not isolated from each other, then a single drive voltage supply is sufficient.

The optical or RF isolated gate drivers (the ones without a secondary power supply) have a low current output, and will increase your switching loss, especially if you have a 10A load.
 
A couple of things are not clear.

1. Do all 96 circuits have to be isolated from each other, or only from the system supplying the on/off signals?

2. To be clear - are there 96 separate, fully isolated voltage sources, each capable of 10 A?

3. To be even more clear, do any of the 96 output circuits share a common ground, either with each other or with the controlling system?

4. What are the minimum On and Off times for the worst-case switched output?

5. What are the maximum turn-on / turn-off ramp times you can tolerate?

ak
 
Last edited:
...please may i update...?
We are seeking an Analog switch (preferably multiple in a chip), which can switch voltages which are outside its Vcc range.......or......the Chip have a wide 0_Vdd range so that voltages up to 40V can be switched.

The attached is what we wish to do. We need to turn on a FET that is not referenced to ground. And there are many such FETs, so we want to use an IC full of Analog switchs.

We preferably need to switch voltages up to 40V
 

Attachments

  • Analog switch to turn NFET ON.asc
    2.2 KB · Views: 162
  • Analog switch to turn NFET ON.pdf
    94.2 KB · Views: 173
1. What happened to the back-to-back FETs?

2. In your schematic, everything is referenced to one common GND. The gate drive voltage (V2?) is not isolated from V3, M1, etc. Does this represent the actual system?

3. Everything in the schematic is functionally digital. That is, the analog switch components SW1 and SW2 are switching constant voltage power sources, not audio, video, etc. Also, M1 is either on (enhanced, saturated, whatever) or off. If this represents the real system, then you don't need any true analog switches, just high voltage switch transistors. So -- which is it?

I reviewed the questions in post #13. No changes, still need answers.

ak
 
Last edited:
It's always good when the OP answers questions but don't hold your breath!

Mike.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top