Onix OA21 Carbon resistor?

[fatmarley]

I blew a channel in my Onix and finally decided to repair it. I found that they used a carbon resistor in one position and I don't understand why. Here's the circuit diagram:



It's the 3.47k resistor circled in red. I need to replace it because it's drifted to 3.93k and wondered if it would degrade the sound if I used a metal film?

Doing a Google image search reveals that they always used a carbon resistor in that position (directly in front of the fuse).

 

[AndrewT]

I think your sch is not correct.
The diode across the ZTX653 looks as though it is meant to be a CCS.
I don't know what the two ZTX753 are doing.

 

[JonSnell Electronic]

If it was 3k3 it wouldn't matter as it is not a critical value. I expect the original was carbon film. Metal film or metal oxide is ideal.
The two ZTX753 are a long tail pair driven by the BC239s.

 

[JonSnell Electronic]

What an awful circuit, too much to go wrong.

 

[fatmarley]

What an awful circuit, too much to go wrong.

Got any suggestions for improvements?

 

[JonSnell Electronic]

Probably best to re design it. That circuit is not the same OA21 that I found on Google; https://www.google.co.uk/search?q=o...2&ved=0ahUKEwjEm6SAyqTMAhUpIsAKHWPQD9IQsAQIGw
Knowing Google it is probably wrong. Ah, I see where your drawing came from.

 

[Ian Finch]

@Jon - Is this the schematic you refer to ? Nothing like the OP sketch and looks to be for the Ion Obelisk, though.
Can you post the actual URL of the Onix schematic you found?

[IMGDEAD]http://www.growthconsulting.frost.com/web/images.nsf/0/16D74AD85FE2F98F8025722800385D59/$File/Obelisk_power_amp.gif[/IMGDEAD]

 

[sgrossklass]

The diode across the ZTX653 looks as though it is meant to be a CCS.
I don't know what the two ZTX753 are doing.

Form a second LTP stage, essentially. And then the current in one leg is mirrored out using the ZTX653.

I don't see too much inherently wrong with it.

BTW, 3.47k seems like a very odd value for a carbon (film) resistor - due to typical tolerances, I would expect one of those to be E24 series at best, i.e. either 3k3 or 3k6. 3k48 is an E48 or higher value.

 

[fatmarley]

Does anyone know if I should match the transistors in the Onix (hfe vbe?)

 

[Ian Finch]

Your schematic seems to have been lost to a timed-out hosting service somewhere.

Going on a recollection that it had a balanced, push-pull Voltage Amplifier stage, I would try to match at least Vbe of the input stage Long Tailed Pair Transistors. You will need a lot of parts from the same batch to get a good match of Vbe and Hfe. It's not always of much benefit either. ZTX type transistors are not cheap so you need to think about matching and using what you have before lashing out on quantities of 50p transistors. Matching Vbe though, is simple to do with even the diode test on a digital meter but it's temperature sensitive so use clips, hook probes or sockets on a breadboard etc. but don't touch them whilst measuring.

If you don't understand the terms here, Google is your friend and helps make a good start to learning some electronic design basics by yourself. More to the point, unless the semis are blown, there's no point in replacing them - not at this stage of getting it to work at least. To check them, first measure resistance between C and E and it should be near infinite out of circuit but most commonly, will be shorted if bad and you can easily measure that in-circuit (unpowerered) too.

The place to start with amplifiers that have "blown" is where the damage is most likely to occur - at least 90% of the time amps fail. That's the output stage where the larger transistors on the heatsink and their driver transistors live. Check these for C-E shorts and all Vbe diode voltages when powered, first thing. It's wise to fit a lightbulb limiter in series with the mains supply (active) lead, to prevent further damage when checking blown amps, too. Not for the incompetent with 230V but necessary

This site is also a goldmine of design info, repair procedure and good projects of all descriptions, if you want to explore: Elliott Sound Products - The Audio Pages (Main Index)

 

[fatmarley]

I was asked to remove the schematic because a friend got help from the designer (Tony Brady) to write it out but Tony didn't give permission to publish it, and he thought it may be morally wrong to do so, which I can understand. Tony has been a great help in the past.

Anyway -The Peak DCA75 Pro just turned up so I measured the BC239C Long Tailed Pair transistors in the good channel - Drum roll . .. . . . HFE 486 and VBE 0.78 (the tranny nearest the input), and the tranny connected to the feedback loop measures, HFE 661 and VBE 0.78. That's quite a large difference (36%).


On the other channel the tranny nearest the input measures HFE 445 and VBE 0.77, but the tranny connected to the feedback loop is showing as a diode, so obviously faulty .

I know in other amps that the LTP trannys should have a mismatch of 10% to work properly and I wonder if that's the case here? (but maybe with a higher percentage?)

 

[Ian Finch]

If you are breaking a confidentiality understanding in seeking help here, then it's probably Mr Brady who needs to answer more of your your questions privately.

I think that after 30 years have passed with the product out there for others to copy freely if they wished, that keeping the design cloaked in obscurity now is rather morbid. I would be happier actually, that someone took an interest in my design some 30 years after release and showed to others what I achieved, maybe recreated it in DIY form, even with a few indulgent touches that modern parts and technology allow.

However, we can't discuss, advise or assist you unless we have a schematic to refer to, so this thread is now pointless. According to your measurements and as I thought though, only the LTP Vbe values would be matched because the cost of manually sorting, matching and disposing of any rejects for the rest would be prohibitive and unnecessary.

I support the right of designers and manufacturers to make it hard to copy their products for as long as they gain income from sales or service. When this ceases though, it's plain that owners (the public), have to take service matters and product information into their own hands. Designers probably know that they won't get a bean for their earlier work once the product life is over, competitors have examined, copied and filed it for their records and the best option now is just to publish and at least get wider recognition for it.

 

[AndrewT]

If the designer does not give permission, then the designer MUST give customer support.

But pester him enough and permission may be granted.

 

[fatmarley]

To be fair to Tony, he may be more than happy for me to share the circuit diagram. I'd just rather ask his permission first. I sent him a message today but he hasn't got back to me yet (he works for Exposure now).

If he gives permission, I'll upload it again...

 

[fatmarley]

Just measured the power transistors.

MJ15003 HFE 388 VBE 0.57v
MJ15004 HFE 264 VBE 0.56v

MJ15003 FAULTY
MJ15004 HFE 276 VBE 0.56v

 

[Ian Finch]

Those Hfe numbers are silly for power transistors. When a more appropriate test current is used they will be much lower, like the product data sheet shows, p2: http://www.onsemi.com/pub_link/Collateral/MJ15003-D.PDF
My DCA75 gives much the same results and is no better than ₤5 Ebay testers there because of the small test current available - useless for big-chip power transistors. It's enough though, for go/no go testing.

Since there was a parting of the ways between Brady and newly formed Onix Audio as long ago as 1985, perhaps the permission to publish isn't just his to give?
https://en.wikipedia.org/wiki/Onix_Audio
Onix DNA | Home

 

[fatmarley]

Those Hfe numbers are silly for power transistors. When a more appropriate test current is used they will be much lower, like the product data sheet shows, p2: http://www.onsemi.com/pub_link/Collateral/MJ15003-D.PDF
My DCA75 gives much the same results and is no better than ₤5 Ebay testers there because of the small test current available - useless for big-chip power transistors. It's enough though, for go/no go testing.

Since there was a parting of the ways between Brady and newly formed Onix Audio as long ago as 1985, perhaps the permission to publish isn't just his to give?
https://en.wikipedia.org/wiki/Onix_Audio
Onix DNA | Home

Ah, that explains it. Was scratching my head trying to work out why my measurements were so different to the data sheet.

 

[fatmarley]

Just got a replay from Tony brady - They didn't match the input transistors.

 

[Ian Finch]

Fair enough. Probably though, in the lots they purchased, there was a closer than random distribution of Hfe. Some amplifiers could well have had matching with a few percent as you found but who's to know how bad the match could also be, yet still sound fine?

Worth experimenting? - You bet and please post what you find here

 

[fatmarley]

Fair enough. Probably though, in the lots they purchased, there was a closer than random distribution of Hfe. Some amplifiers could well have had matching with a few percent as you found but who's to know how bad the match could also be, yet still sound fine?

Worth experimenting? - You bet and please post what you find here

The blown channel now has perfectly matched HFE and VBE. The other is left as it was. Just setting the bias now...