Short circuit output protection scheme

[tool49]

Hi there!

I've been playing around with a new amp design and while researching I noticed that most emitter follower amplifier with output short circuit monitor the voltage drop in the emitter resistor of the output stage.

While this might be alright for an amp with very little limitation required, when more effect is needed, the value of those resistors need to be increased. Now this seems somewhat not very desirable. It means more wasted power in the emitter resistors and an increase in Zout of the amp.

I was wondering if anyone had tried to instead monitor the drivers. In the emitter follower output stage, the drivers are usually also linked to out with resistors. These resistors are usually of higher value thus giving more swing for the protection to kick in faster.

Now, I might just be looking at the problem with my nose sticked to it and not be seing the big picture. Do you guys see any disadvantage at using the driver's emitter resistors? What am I missing?

Thanks,
Sébastien

 

[Nigel Goodwin]

The important thing you need to measure is the current through the output transistors, it's those that provide all the high current. Monitoring the drivers is only protecting the drivers, and not the output transistors - the current relationship between the drivers and output devices isn't going to be constant.

I don't quite understand your concern though?

While this might be alright for an amp with very little limitation required, when more effect is needed, the value of those resistors need to be increased.

The value of the resistors is dependent on the output current you wish to monitor, a smaller amp requires larger resistors, in order to drop the same voltage.

The output impedance of the amplifier shouldn't be affected very much, as the resistors are inside the negative fedback loop.

 

[tool49]

Hi Nigel,

the current relationship between the drivers and output devices isn't going to be constant.

I tought it would be since the ratio of the emitter resistors is constant. I shall have to think about this one some more.

The value of the resistors is dependent on the output current you wish to monitor, a smaller amp requires larger resistors, in order to drop the same voltage.

I'm trying to build something around 500Wrms@4ohm. With 0.1ohm I can't get the protection to kick in early enough. I need at least 0.33ohm which dissipate quite a lot of power and also diminish the amp's ability to swing within a volt or so of the supply.

I'm going back to my simulation!

Thanks,
Sébastien

 

[johnnyx]

Use a potential divider across the output transistor to raise the voltage a bit. ie sense the v drop across the 0.1 ohm resistor + a bit of the voltage across the transistor. In this way, when the output voltage is low (eg. with a short circuit load) the current limit is at one level to prevent overheating, and when the output voltage is high (peak output level) the current limit level is increased because less is dropped across the output transistor, and allows higher peak currents.
Hope this makes sense, hard to do without a picture. You will see this scheme in other designs though, eg Doug Self.

 

[AndrewT]

Hi Tool,
send me an Email address if you can use excel.

 

[tool49]

Amplifier knowledge required!

Hey guys!

I hope I got what you meant right johnnyx. Anyway this is what I came up with from what you had written as a protection scheme. (R28-R36 + D1-D4 + C8-C9 + Q8-Q9)

An externally hosted image should be here but it was not working when we last tested it.

I works quite good at limiting the current output.

The amp simulates quite good at low freq but at high frequency, it is horrible!

Look:
100Hz

An externally hosted image should be here but it was not working when we last tested it.

10KHz

An externally hosted image should be here but it was not working when we last tested it.

How would I go around to improving the high frequency response? (I suspect insufficient slew rate in the input stage but I might be wrong!)

Thanks,
Sébastien

 

[AndrewT]

Hi,
email sent.
Try reducing or eliminating the Miller compensation cap. I am not a fan of this stability sorter.
Have you got balance in the LTP pairs.
33k gives about 2.5mA tail current but 1.25mA through the 1k5 LTP load gives 1.93V. This seems very high and would require a VAS Iq =8.5mA to achieve LTP balance.
Add a capacitor to q8 base to output and q9 base to output. Just in case the sim is protecting the outputs on fast slopes.
r6 & r7 are injecting PSU ripple into the non inverting input. Have a look at KSA50 to see how Krell did it.