Salas shunt help

[agithegreat]

Ok I have myself a DAC here which I built up from the group buy a few years ago:
OLD THREAD DAC End by Andrea Ciuffoli

It uses a salas shunt supply. Now, only 1 of the 3 shunts has failed.

Refering to this:

It's the bottom one which has gone.
Well it still works as such. I still can vary the voltage (using the 100r trim) and there is VERY little noise on the output. However the first 3 LED's do not light up. The ones between G and S on Q1.

Also the resistors, R1 (marked as 33r) get very hot. I mean they smell and get far too hot to touch.
Now as the thread progresses it's noticed that 33r is too much really for the TL431's to regulate. So I knocked this down to 13r (2x26r).
I checked the resistance across and I still get 13r.


What shall I check now? Why would the first 3 LED's not light? I guess it's either the 9240 shorted or the 2sk170 blown, right?

[merlin el mago]

Yes change first the LED's to assure the CCS is working OK & let us know?

[agithegreat]

Alright, i'll do that.

I get ~1.8v drop across each working LED (measured on the other shunt) Across the none working shunt I get ~0.25v across each LED.

I'll report back with new LED's.

EDIT:
Also I get 13.4v across all the LED's (From R1-Q5)
On the working one I get ~5v.
If I put in new LED's wont they blow instantly with 13v across them?

[AndrewT]

check your input CCS jFET & mosFET.
add in a high power 10r in series between the shunt PCB and the transformer supply.
check voltage around the input CCS & FET as you increase the input voltage.

If the Fet operates with 3.5Vgs and the LEDs drop 5.6V then the Resistor should drop (5.6-3.5) = 2.1V This allows you to check the CCS current and you will find this is grossly low if the input voltage is low.
All the devices vary in their actual parameters from average. You do need to check as you build.

I would suggest that the recommended input voltage be increased from 10-12Vdc to 12V-15Vdc. This will make the input FET run hotter.

[agithegreat]

Ok.

Vgs of Q1 is 5.3v.
On the working shunt it's 3.7v

Input voltage is 13.4vdc. Is this still too low, you think?

V drop across LED's on blown reg is 1.3v
V drop across R1 is 8.2v! R1 in my case is 13r though... (2x26r)

The other regs work fine. Also I have removed the TL431's off the DAC board now as I think they were what was causing the problem. The new rev of this setup doesn't have them in as many people wondered why they were there, as the shunt can provide clean enough voltage. It was just a vicious circle as you needed a high enough to them (5.6v or so iirc), but this shunt couldn't keep up so r1 was lowered, this then increased current flow and in turn resulted in it getting too hot. Now something has blown because of this.

[AndrewT]

What is the voltage across the CCS resistor?
Does that voltage change by much if you lower the input voltage by 1V or by 2V. Use that power resistor to experiment and it will tell you a lot about how your shunt works.

The input you have measured was at that moment. It varies continuously and over a surprisingly large range. That's why I suggested the higher input voltage. Your measured input voltage fits inside my recommended range.

[AndrewT]

You cannot blow a Shunt Regulator that has been set up correctly. It is inherently short circuit proof. One of the tests I carry out is to insert maximum supply voltage and to short the output to ground. This forces the input mosFET to maximum dissipation. My "pass test" is that on removing the long term short that the shunt operates correctly having suffered no heat damage. Some builders may consider this over the top engineering. You decide.

You can modify the Shunt regulator to ensure it will eventually fail because of gross incompetance.

[Salas]

Those pcbs had small on board sinks as I recall. Could the input 9240 had blown due to increased CCS current? Is it getting hot at all now? Prime suspect.

[agithegreat]

I did a ninja edit whilst you were posting with v drop across R1.
But to re-iterate it's 8.2v! R1 = 13r
So thats like what 0.63amp?

I don't have any handy dandy way to change the voltage input .

[AndrewT]

Post 4

insert a power resistor in the feed !!!!