Repair help needed - Creek Destiny

Hello everyone,

I have a Creek Destiny integrated amplifier (version 1) which has burned twice the output transistors. Here 's a brief story:

- on the first week of purchase it oscillated and burned the mosfets, repaired by the local dealer (replaced transistors without investigating the cause of the oscillation). Note that my speakers are inductive up to 48KHz where I can measure.
- after a few hours of playing (but years later and out of warranty, i don't use it often) it went to protection mode. I replaced four blown electolytics (at the pre-amp stage) and the amp came back to normal operation, but minutes later it oscillated again and burned the mosfets of the left channel.
- I replaced the mosfets ( 76639P ) and I got sound from both channels. I tried to check/adjust the bias. On the board it is printed "adjust vr1 to 21.5mV accross jp4 at idle" so I checked the voltage across JP4 on the right channel (the one that did not had a problem) and it was 8mV (which seems suspicious, shouldn't it be at the suggested value?) so I adjusted it to 21.5mV. I tried to do the same with the left channel (the one which I replaced the mosfets) but the voltage drop was 0.2mV and does not change at all when i turn the related trimmer.

Unfortunately I can't find a schematic so this is a quite difficult task for me. Any ideas?
 

Attachments

  • DSC07833_.jpg
    DSC07833_.jpg
    249.8 KB · Views: 1,005
Last edited:
I was comparing the 5350's schematic with destiny's pcb and as far as i can tell it has the same components so the circuits must have very slight differences. What else could I test except from comparing voltages between left and right channel on the corresponding components? Does anyone understand what could be wrong and the voltage drop across R28 is 0.2mV no matter what the trimmer's adjustment is?
 
Administrator
Joined 2007
Paid Member
That's one of the strangest circuits I've ever seen.

Can you just confirm the amp actually plays on both channels OK as it stands ?

If so, then I'm wondering if the non adjustable bias is more of a problem with the characteristics of the replaced transistors.

Anyhow, it looks as though the voltage between the drains of Q12 and Q14 are what is going to determine the current. I would compare those voltages, one channel to other first, in order to get some idea of what is going on. You must measure between those points and not to them individually.
 
Thank you very much for the reply. Yes the amp plays on both channels just fine. I didn't dare to turn up the volume or let it play for much time though, I am afraid to stress-test it until I resolve the bias problem. I will follow your advice and focus on Q12 & Q14 :)
 
Can you just confirm the amp actually plays on both channels OK as it stands ?

If so, then I'm wondering if the non adjustable bias is more of a problem with the characteristics of the replaced transistors.

Hi, your hunch must be right on target. The amp plays on both channels just fine. The spare 76639p I purchased were bought from chinese seller since they are obsolete and could not be found in a respected distributor. I measured them and compared to the original ones (the right channel fortunately was not burned). While in off state the original ones have source-->drain diode measuring of 525 while the "chinese" had 510. But while in ON state, the originals have nealry zero measuring 001 while the chinese give me a source to drain diode measuring of 070-100. If I understand well this shows the ability of the mosfet to conduct, and if this translates to a quite high Rds-on then this could explain the bias adjustment problem, could it ?

The only close equivalents I could find are STP40NF10L and I don't see another option but to replace them with these.

edit: According to this article, the circuit demands "logic level" mosfets which means low Vgs-th (correct me if I am wrong, I am just finding out what these mean) and STP40NF10L must be ok on this (min 1V max 2.5V while the 76639p has 1-3V ), are they?

I also spotted a flatulent smd capacitor, C20 near the TL074, I think its 100nF ceramic bypass capacitor. Do smd caps fail and get flatulent like electolytics? Could this cause an oscillation?
 

Attachments

  • DSC07833__.jpg
    DSC07833__.jpg
    450.3 KB · Views: 655
Last edited:
Administrator
Joined 2007
Paid Member
Its pretty rare (very very :)) for an SMD cap to fail in low voltage circuitry such as around an opamp. I'd bet its just fine.

Rds and Vgs are two different things really. A low Rds means that the FET appears as almost a short when turned fully on... and thats what is needed for power control using FET's as switches to minimise losses. But low or high Rds, that in itself isn't the issue here.

In an amplifier like this, the FET's are not used in that way (as switches) but are used in the "linear" part of the transfer characteristic.

Vgs is a measure of how the D-S current varies with a change in G-S voltage. And I think that is where your problem lies, the circuit has a very limited adjustment range and its outside of the Vgs of the replacement devices.

I'll have another look at the circuit later today (its on a different PC) and see if anything can altered.
 
yes I did measure the voltages:

I dont have my notes with me (I am at work now) but if I am not mistaken Q14 has 3.8V (Q14 source to ground) while Q114 (the other channel) has 4 volts.

Can't remember Q12, I will let you know as soon as I return home. Thanks again for being helpful
 
Administrator
Joined 2007
Paid Member
yes I did measure the voltages:

I dont have my notes with me (I am at work now) but if I am not mistaken Q14 has 3.8V (Q14 source to ground) while Q114 (the other channel) has 4 volts.

Can't remember Q12, I will let you know as soon as I return home. Thanks again for being helpful

No problem :) but remember its the voltage between those points thats important, not the voltage from each to ground.
 
Vgs is a measure of how the D-S current varies with a change in G-S voltage. And I think that is where your problem lies, the circuit has a very limited adjustment range and its outside of the Vgs of the replacement devices.

I'll have another look at the circuit later today (its on a different PC) and see if anything can altered.

this enhances my intention to replace all mosfets with STP40NF10L, because now I have questionable origin and specs mosfets on the left channel and factory mosfets on the right channel.
 
Administrator
Joined 2007
Paid Member
this enhances my intention to replace all mosfets with STP40NF10L, because now I have questionable origin and specs mosfets on the left channel and factory mosfets on the right channel.

I can understand your thinking on that. I would still measure and confirm the voltage I mentioned though.

Remember there can be a wide tolerance on Vgs. An IRF240 (really common power device) has a min of 2 volts and a max of 4 volts quoted as thresholds in the data sheet. Maybe Creek used selected devices of a certain low Vgs.
 
Vd_q14 - Vd_q12 = 52.8 volts

Vd_q12 = -49V (black probe to ground)
Vd_q14 = 3.8V

Vd_q114 - Vd_q112 = 54 volts (on the good channel)

if this makes any difference, Q12 & Q12a are zvp2120 on the destiny instead of zvp3310 mentioned on the 5350's schematic.
 
Last edited:
The only close equivalents I could find are STP40NF10L and I don't see another option but to replace them with these.

edit: According to this article, the circuit demands "logic level" mosfets which means low Vgs-th (correct me if I am wrong, I am just finding out what these mean) and STP40NF10L must be ok on this (min 1V max 2.5V while the 76639p has 1-3V ), are they?

You are right, output mosfets have to be logic level...one of the good candidate is Siemens/Infineon SPI47N10.
 
Administrator
Joined 2007
Paid Member
Vd_q14 - Vd_q12 = 52.8 volts

Vd_q12 = -49V (black probe to ground)
Vd_q14 = 3.8V

Vd_q114 - Vd_q112 = 54 volts (on the good channel)

if this makes any difference, Q12 & Q12a are zvp2120 on the destiny instead of zvp3310 mentioned on the 5350's schematic.

Easy one first. C20 is just a rail bypass. If the 15 volts is present and correct across it then its going to be OK.

Your voltage readings. It was this voltage I was after. If getting FET's with the correct Vgs range is an issue then it still might be possible to alter the circuitry to suit but it would have to be done carefully and with a bulb tester in place to limit current. Adding another diode in series with D8 (circled) would add another 0.65 or so to the Vgs range available.
 

Attachments

  • Capture.JPG
    Capture.JPG
    149.9 KB · Views: 372
Administrator
Joined 2007
Paid Member
yes the voltage in your drawing is 52.8 volts

OK... sorry that's my fault, its been one of those days lol :) I'm losing the plot a little bit with this one, forgetting its a quasi complementary stage... its such a strange circuit to visualise.

I still think the problem is down to the different Vgs characteristic of the replacements. Do you think you can get better devices than you are using now ? If not then it still might be possible to juggle this around a little to achieve the desired result with what you have but its all an unknown and would require care and a lot of thought.
 
I have ordered STP40NF10L which are the closest equivalents to the factory mosfets (HUFA76639P3) from Mouser.com and I hope I will have them in my hands in a few days.

I would prefer to try a solution like this first, because i will have both channels identical, if this doesn't work as expected then I will have to try your modification advice

Another option would be to risk again and buy HUFA76639P3 from ebay but I doubt if they are originals.