Electrical isolation

[Alex]

Hi,

I currently work on a circuit designed to control the output power of a DC/DC
converter. I heard that insulating the control circuit from the power circuit
could reduce the noise in the controller and protect the control circuit from
any incident that can occur in the power stage. However, I'd need to isolate two
analog inputs (voltage and current), the MOSFET's gate signals and the power
supply (??).

I'd want to know if anyone has an idea about the usefulness of a complete
electrical isolation. If it's necessary, how can I isolate each of the signals
and how do I manage the power supplies and the grounds?

Thanks,

Alex

 

[R.Legg]

Hi,

I currently work on a circuit designed to control the output power of a DC/DC
converter. I heard that insulating the control circuit from the power circuit
could reduce the noise in the controller and protect the control circuit from
any incident that can occur in the power stage.

The reduction in potential damage is only important in circuits
designed for repairability - if this is high volume commercial,
there's probably no point. Noise is a problem best tackled directly,
both at the source and the recieving end.

There are also other methods available to protect control circuitry
when directly coupled
- impedance-limiting higher-powered sense lines,
- zener-limiting voltage amplitudes,
- resistively-decoupling supply voltage partitions,
- the conscious application of dedicated sacrificial circuits (ie
fusible resistors, fusible zeners, coupling caps with suitable voltage
rating, magnetic parts with suitable volt-seconds).

However, I'd need to isolate two
analog inputs (voltage and current), the MOSFET's gate signals and the power
supply (??).

The current is cheaply monitored using commodity current transformers.
Voltage can be sensed on the relevant secondary waveform that reflects
the primary voltage of interest, by low-power peak detection.
Mosfets can be driven by transformers also. If these function only to
increase noise margins or improve repairability, then they can
obviously be employed within the directly-coupled circuitry on the
primary, without resorting to total secondary-side control.

The main problem with secondary-side control is developing start-up
power. This is usually the biggest budget item working against the
configuration in an economical circuit. There are a few ways around
it, all of them with their own limitations.

I'd want to know if anyone has an idea about the usefulness of a complete
electrical isolation. If it's necessary, how can I isolate each of the signals
and how do I manage the power supplies and the grounds?

The bottom line is $ vs function. You have to know how to meet your
noise and repairability requirements without isolation first, before
you can responsibly compare the cost of implementing 'complete'
isolation as an alternative.

RL