advertisement


HackerNap failed

I replaced all the semiconductors and just the resistors that had gone open circuit or burnt, brought supply up slowly on a variac and got that satisfying magic click from the Velleman relay. Quite chuffed by now I went to set the bias, just about this time I buggered it up by catching the meter probe on 0v and the output emitter resistor.

So I replaced all the semiconductors again except the FETs and one damaged emitter resistor R27, this time I have been defeated by having some output volts.
Output -45V
Tr9 base -44.6V
Tr10 base -47.4V
Other than ordering a further set of parts I am at a loss and needing some diagnosis help.

Thanks
Tony

On the plus side I have now fitted emitter resistor test terminals
 
The Naim protection circuit could be fitted, but that did have an effect on the music.
Click to expand...
Point of order, that circuit is only SOA protection - deals with secondary breakdown in transistors - and will not catch or manage the effect of driver or output stage failure. Many, many production amps at all scales feature the same, or variations on the theme BTW; it's just good housekeeping.

However if you really want DC/offset/failure protection - esp. on a DIY effort - something like the Velleman relay board is the way to do it.


PS detecting oscillation in a way you could protect against means you could probably fix it in the first place ;)
 
Tony - I don't have the schematic to hand but since the output has gone hard to -ve rail that looks like the 'upper' or +ve rail half of the OP stage has gone (unreliable at least) - check this emitter resistor for connectivity too.
 
Thanks Martin, emitter resistors have continuity, in fact I have continuity from the upper emitter to the lower collector.
I can increase bias up to say 4mV in the upper emitter resistor when its 3.4mV in the lower.
The upper transistor is warm, the lower is cold.

I seem to have -22 volt (negative 22 volt) on the output of the upper (+tve) front end FET, I cant actually get the probe to the FET pins but can get on the upper half of R7 so I measured at R7.
I think this should be around +46 volt.

Looking at the upper front end again I have 80 volts across the upper FET and 9 volts across the lower FET.
To my simple mind it seems as though the upper front end FET is not conducting.
Could I just jumper the positive front end across the FET to TO R7 to prove this?

Tony

Edit
Seemed obvious this morning to compare the upper front end on my other HackerNAP. Sometimes I cant see the wood for the trees.
So right now it looks like the FET has failed.
 
Changed the FET and its all rock n roll now folks :)

Martin gave me the clue, until then I was chasing faults on the lower half, duhh I must get to dips the class B amps.

Tested a FET for the first time
N-Channel. I guess P channel is just the opposite procedure.
First connect dmm red lead to drain and black on the source.(meter should be on diode test mode)
in first case drain to source will be open.
Then connect meter red probe to gate and black to source and again connect the red probe to drain and black with source.
Now it will show the short circuit. After this short the gate and source with some wire now again drain to source will show open circuit.
If the above works the FET is ok.

Many thanks for your help
Tony
 
Glad to hear you're up and running again. Do you know what the root cause of the problem was?
 
Hi Carl,
Yep root cause, = stupidity,
Not letting the capacitor banks to fully discharge before swapping pre amps seems to have led to some nasty oscillation which took out most the lower output stage.
Quite nasty, resistors with holes in them.

Second time it failed was just after the repair when setting the bias I shorted the upper emitter to the adjacent ground connector, I have since fitted some safer test pins to the emitter resisters. Would be a nice board revision?
As knew instantly what I had done I shut down and damage was limited to one output transistor, its driver and the FET. Just took a while to track down the faulty FET as I am limited in diagnosis skills.

Glad your here though as the bias has to be finally set, at the moment its 5mV over 0R22.
What is your recommended bias?
Also emitter voltage is different on the two emitter resistors giving 5mV on the upper and 4.1mV on the lower, I guess its just resistor tolerance. some hundredths of a ohm?
Should I worry.

Thanks
Tony
 
Ugh, sucks. Maybe think about fitting the resistor/LED bleeders to help discharge the cap banks.

The voltages sound fine, and for bias I'd aim for the value specified in the build manual - around 3.5mV.

Cheers,
Carl
 
Thanks, its fitted with led bleeders but they are only set at 12mA so you have to wait a while, I wont get caught out again though, thinking of wiring them to the front panel as a visual also.
 
I must admit Tony its a good point with 24 Kendiels in there (like grey hound's fighting to get out of the trap) you realy do have to be carefull about connecting up the boards in particular.
I made a little resiter drain kit up for mine
I wouldn't have thought just connecting up a source would cause a failure if you have input caps but I know you can get quite a pop or crack if you have the gain up when connecting / dissconnecting to the pre amp, I allways just knock the pre amp switch to a position with no source when I am doing that.
Any way glad to hear its making music again;)
Alan
 
Bemused said:
Second time it failed was just after the repair when setting the bias I shorted the upper emitter to the adjacent ground connector, I have since fitted some safer test pins to the emitter resisters. Would be a nice board revision?
As knew instantly what I had done I shut down and damage was limited to one output transistor, its driver and the FET. Just took a while to track down the faulty FET as I am limited in diagnosis skills.
Click to expand...

I've done that too on my clones and taken out the output devices as well. Now when I set the bias I use the Avondale method and put my test meter in the positive line and read the current draw. It's so much easier as you can simply leave it all connected up and monitor it over half an hour or so. Don't forget to make sure those caps are fully discharged before putting everything back together after though ;)
 
Chops54 said:
...Now when I set the bias I use the Avondale method and put my test meter in the positive line and read the current draw. It's so much easier as you can simply leave it all connected up and monitor it over half an hour or so....
Click to expand...

I'm not sure that this method is applicable to the HackerNAPs because they use a separate front-end supply, which means the current drawn is actually quite different than you'd expect.

Unless you've done the necessary calculations I suggest measuring at the emitter resistors for the HackerNAPs or any other NAP derivative amp with a separate front-end supply.
 
My NCC200s have seperate front end supplies. I remove the positive feed to the output section and insert my meter between the lead and the amp. As the HackerNap is more or less a copy of the NCC200 I can't see why this method won't work. Sure beats waving your test leads around in the middle of all those components ;)
 
Chops54 said:
My NCC200s have seperate front end supplies. I remove the positive feed to the output section and insert my meter between the lead and the amp. As the HackerNap is more or less a copy of the NCC200 I can't see why this method won't work. Sure beats waving your test leads around in the middle of all those components ;)
Click to expand...

The front-end is a constant current circuit, which means that unless you know exactly how much it's drawing and you're subtracting that value from the ~38mA being measured in the +ve rail, you could be setting the bias inaccurately.

From memory I don't know what the front-end draws. It could be small enough to be negligable/irrelevant, but it's probably prudent to check. Edit: actually I think it's the -ve rail that has the CCS. Will need to go back and check.
 
The current draw of the front end is approx 11 mA per channel. With no signal both rails draw this constant current (any small differences between rails are due to gain imbalances in the output stage and LTP) The negative rail of the front end is truly constant current even when there's a driving signal. The front end positive rail has small signal related AC variations on top of this DC constant current.

If 11 mA goes to the front end that leaves 27mA for the output stage, all of which goes through the 0.22 ohm resistors, which gives us 5.9mV across each of these. To my mind this is a good way to set the bias up without poking around a live circuit. I believe it is what LesW recommends.
 
Thanks John, some nice concise figures, seems a safer approach and negates ant emitter resistor tolerance error.

Tony
 
I have had a look at Martin Clarks notes again on bias setting and also a few threads on here, it sure can give you a headache...

I have over the last day increased the bias from 3.5mv to 5.6mV.
Blimey what a game, lid on / lid off / heating on / heating off / speaker connected / speaker disconnected / breathing / holding breath, it all alters the voltage.

Source and pre connected, volume set to zero and source stopped, speakers connected.
There is 16mV on the output terminals which is pushing 0.16mA into the speaker.


I will try and table some voltages below

Upper_______________________Lower
5.6____Speaker connected_____3.2
4.8____Speaker disconnected___3.7

Does this make any sense?

I think I will try Johns current method next.
Should the speakers be connected during the test or not?

Tony
 
Safest to not have speakers connected, and won't change the bias setting directly (does indirectly by making the amp run warmer at high volumes).
 
Set the bias with the speakers disconnected. There will always be a small amount of dc on the output. 16mV is a pretty good result. If dc on the output bothers you then fit some loudspeaker protection boards.
 
You must log in or register to reply here.


advertisement


Back
Top