Weird HP Power Supply Circuit

[Gregory]

Hello!

I`m trying to repair a HP 35660A power supply that failed.

As I was not able to find the schematic, I`m doing a bit o reverse enginnering to see how the power supply works.

On this power supply I came across a power driver design that I never saw before, this is the LTSpice print that I simulated.
(I used a resistor in place of the main transformer only for simulation purposes)


Q1 is the main switching transistor.
The weird thing is that M1 is a FET that is in series with Q1! In between them it has a small transformer that appears to be a boostrap transformer to inject base current on Q1.

The zeners are a small voltage supply that bias the base of Q1 with a small current to start the conduction.

So, Q1 is controlled by its emitter by M1 and when it starts to conduct current the bootstrap transformer reforces the base current. Very weird.

Why do you think they used this design ?
I can't see any advantges of this counblesome design.

Thanks!

[Cliff Matthews]

Check the description text below EEVblog video #523 - I think you'll find the resources you need.
* Spoke too soon   I re-checked the svc man PDF and you can find it here.

[Gregory]

Thank you for your repply but the document don`t contain details abou the power supply!

[Andy Watson]

Weird! I'm going to guess that it is a slight variation on a cascode circuit, i.e. it is an attempt to reduce the miller-effect around M1.

Another possibility is that M1 is only good for 100V whereas Q1 is good for 1000V - perhaps there was an ecconimic reason for using a Mosfet and the IRF530 was the best they could obtain at the time.

[NiHaoMike]

That's called an "ESBT", Emitter Switched Bipolar Transistor.

[Gregory]

That's called an "ESBT", Emitter Switched Bipolar Transistor.

Thank you!

So, the advantages are:

- It can use a very high voltage BJT and a low voltage MOSFET
- The miller effect is minimized because the MOSFET only see a small percentage of the total voltage and when the BJT switchs the current is already cutted by the MOSFET

Do this sounds right ?

[NiHaoMike]

I think the idea is still being used with some kinds of SiC and GaN FETs. The use of high voltage bipolar transistors has faded away a long time ago.

[T3sl4co1l]

It's not very good that there's no outflow path for base current, so it will probably turn off quite slowly, in the process putting quite a lot of voltage across the MOSFET.  Or the diodes carry that current in avalanche.

The BJT drive transformer with positive feedback is a common sight, it's been used in ATX supplies for example.  The combination acts like a dynamic SCR, so the controller needs to produce shorting-mode commutation in order to maintain control.

Here's an example I made, for an LED fixture:



This control uses frequency modulation, and treats the power stage as a one-shot.  Commutation is provided by saturation of the drive transformer (when an inductor or transformer saturates, it effective shorts its terminals, hence sharply removing base current), giving an on-time somewhere around 5µs.

Tim