Advice on replacing transistors in a Rotel

Status
Not open for further replies.
Jump to Latest
L

lorriman

Member
Joined 2011
#1
My amp has DC offset of 86 and 88mv and I'm hoping to get that somewhere manageable but there's no DC offset pot. I've understood, from audiokarma, that the transistors have 'drifted off spec' and probably need replacing along with the caps. So that's the plan. It's a bit of a shame really as it's my first foray in to real hifi and I got myself a dud (unfortunately several months ago, so I can't return it).

Not being an electronics person I wonder if anyone could tell me what transistors I should be buying to replace those in my Rotel. It's the RA-930AX, whose manual is here:

Service Manuals

From reading other posts involving Rotels I hit upon this one:

http://www.diyaudio.com/forums/solid-state/solid...ml#post2238851

Some help at AudioKarma pointed me in the direction of the 2SA970BL as a better choice for a newbie due to complications with fitting the ZTX795A. any opinions on that?

Another issue is that the above thread suggests 'matched' transistors. I've found a few places that sell these but I can't see an option for ordering matched transistors. Am I supposed to order a bunch and match them myself? yikes! I've only a DMM. I'm in the UK, BTW.

Thanks for any help.
 
#2
I should add: one concern I have with the suggested 2SA970BL is that pictures of them on the web don't appear to show a hole through the centre. The transistors in the amp are screwed to a heatsink through their centers. Are the pics inaccurate/generic?
 
#3
Before you blame transistor aging, I find the NFB electrolytics short in older amps. I would replace C605, C606 (100uF 6.3V); they will add a dc offset. Maybe look at C601, C602 (10uF 50V) but they'd make the input gain pot noisy if they are shorted/leaky. Beyond this- the input diff. amp pair as mismatched? - strange to have the same offset in both channels. If it's true, I'd venture you can just swap the high/low hfe parts between the L and R channels to get a better match.

Yes, you buy more transistors than needed and match them. It's a bit of a hassle but part of the craft...
You can get $10 multimeters with hfe measurement, or make a jig using a multimeter to measure beta but don't make mistakes with polarity.

2SA1016F hfe=160-320
 
#4
A little matter - the thread link doesn't work (for me at least)

BTW - +1 to the above post. Those offset readings are highly sus. and not likely due to any transistor "drift". Note that when buying caps, larger voltage ratings are OK - as long as they fit in the available space and lead spacing. IME the lower voltage ratings fail soonest. 'No need for boutique types here, decent industry quality Panasonic, Rubycon, NCC etc. is fine
 
#6
'You sure about item 1 in your list, Fivetide? It might be wise to read up some issues on tantalums before suggesting that one and the 6V rating elcos. Maybe also think about what happens if the output stage fails which often happens when amps get "experimented" with. Usually, you get a fair amount of DC across that feedback cap. Tantalums really don't like that.
 
Last edited:
#9
lorriman, I think you're looking at replacing the output transistors (“...on the heatsink, hole in the middle..”) while others are talking about replacing the small input transistors (Q601,602,605,606; 2SA1016’s) with 2SA970, one fix for the 86mV DC offset you have. This is why the pics don’t match.

For NFB capacitors C605, 606 although they are polarized, put in non-polar electrolytic like Panasonic ECE-A0JN101U
 
#11
Thanks for the help everyone. Interesting reading. 🙂

Here's that link, working this time :

http://www.diyaudio.com/forums/solid-state/169537-rotel-ra-920ax-4.html#post2238851

Am I right in thinking that fivetides' list is perhaps at 'enthusiast level'? I'm really just looking for standard parts so presumably I just need to match what the amp has?

What's the advantage of a non-polar capacitor?

Thanks for that, Prairiemystic. I was getting a touch confused there.

Any recommendations on transistors, folks. The ones mentioned seems to be a little hard to find. I did find them on RC-components but I was hoping to buy from mouser or farnell instead due to other purchases (and postage). When I say hard ot find I mean from trusted sources where I won't end up with a fake.
 
#12
Ian Finch said:
A little matter - the thread link doesn't work (for me at least)

BTW - +1 to the above post. Those offset readings are highly sus. and not likely due to any transistor "drift". Note that when buying caps, larger voltage ratings are OK - as long as they fit in the available space and lead spacing. IME the lower voltage ratings fail soonest. 'No need for boutique types here, decent industry quality Panasonic, Rubycon, NCC etc. is fine
Click to expand...

Just to get a bit of clarity: would you say that the most likely solution is re-capping, or am I looking at something more significant?

What would you assume is the likely cause? Power supply caps? That's my naive guess.
 
#13
There's a decent chance you're measuring the same offset the unit had right off the assembly line. Having guaranteed your meter was zeroed, and you got the polarity right, increasing the current source for the differential slightly, or slightly decreasing the VAS output current source, could restore the offset to zero if hand tweaked. The first thing I'd do is swap the input stage transistor positions and see if you end up with a different offset. That could immediately eliminate questions about differential stage transistor mismatch causing the problem. You'll likely measure 80mV either way.

There was considerable research that determined low voltage (6.3 or whatever) electrolytics have higher signal distortion than 35 or 50 volt units, and that putting a pair in series + to - reduced distortion ever more slightly. I don't have a link to the work on hand, but a 35 volt 220u Panasonic FM cap is only 50 cents. Though not impossible, I don't think capacitor leakage is causing the offset.
 
#14
Agreed - you don't need to change transistor types or electrolytic cap.types. It's just an opportunity for some cap. renewal which will help prolong the life of the amp. since they all deteriorate with tine, temperature and duty. Upping the voltage rating (because you can.... with modern smaller caps) might be just wishful thinking, but doesn't hurt and may slightly improve AC performance.
 
#15
Having looked at the circuit I would say the offset you measure is to be expected. It's not designed for DC precision at all, just look at the base bias currents for a start.

I wouldn't change any semiconductors on this just for the sake of it.
 
#16
Mooly said:
Having looked at the circuit I would say the offset you measure is to be expected. It's not designed for DC precision at all, just look at the base bias currents for a start.
Click to expand...

Sheesh!

Does this mean I have a stinking-pile-of-stuff amp, by design?!?!? I had thought Rotel's were supposed to be half decent.

From what I had understood from AudioKarma, such an offset indicates audible distortion and isn't far off being bad for the speakers. They didn't mention amps designed with DC offset more-or-less built in. I have tested a bunch of amplifiers and most of them have been between 1mv (a 10 year old cheap sony mini) and 25mv (a 'hifi' sony).

The (generalised) rule on audioKarma was : up to 15mv DC offset: just fine, 15-30mv marginal, 30-50mv: not desirable, 50+: something's wrong.

That link is of my amp's little brother and almost confirming your statement, was at 88mv on both channels, and sounded bad. Replacing the transistors, caps and a resistor got it down to 30mv.
 
Last edited:
#17
Andrew Eckhardt said:
There's a decent chance you're measuring the same offset the unit had right off the assembly line.
Click to expand...

There almost seems to be a consensus for that.

Having guaranteed your meter was zeroed, and you got the polarity right, increasing the current source for the differential slightly, or slightly decreasing the VAS output current source, could restore the offset to zero if hand tweaked. The first thing I'd do is swap the input stage transistor positions and see if you end up with a different offset. That could immediately eliminate questions about differential stage transistor mismatch causing the problem. You'll likely measure 80mV either way.
Click to expand...

I can't guarantee the meter though it does measure other amps with normal results (between 1mv and 25mv, so far).

I think adjusting 'differential' or 'VAS output' is beyond my understanding, which is fairly limited. Is this almost as simple as replacing transistors and caps?
 
#18
lorriman said:
Sheesh!

Does this mean I have a stinking-pile-of-stuff amp, by design?!?!? I had thought Rotel's were supposed to be half decent.

From what I had understood from AudioKarma, such an offset indicates audible distortion and isn't far off being bad for the speakers. They didn't mention amps designed with DC offset more-or-less built in. I have tested a bunch of amplifiers and most of them have been between 1mv (a 10 year old cheap sony mini) and 25mv (a 'hifi' sony).

The (generalised) rule on audioKarma was : up to 15mv DC offset: just fine, 15-30mv marginal, 30-50mv: not desirable, 50+: something's wrong.

That link is of my amp's little brother and almost confirming your statement, was at 88mv on both channels, and sounded bad. Replacing the transistors, caps and a resistor got it down to 30mv.
Click to expand...

You have to look at the amp as a whole.
Imbalance of the currents in the LTP does increase distortion but that could be part of the intended design. The distortion produced is 2nd (even harmonic) which is actually pleasing to the ear. That could be intended as Rotels were supposed to have a bit of "human input" to the design.

For correct DC balance the base bias currents need to equal, the input transistors matched somewhat and also the LTP needs the addition of a current mirror to enforce sharing. You could alter the collector load resistor of the first LTP transistor to bring the offset down but that doesn't make it a precision DC amp and the offset will still wander and drift a little with temperature.

The amp is a fairly basic "text book" design, not that there is anything wrong with that. 80 mv offset gives a power dissipation of 0.0008 watts in an 8 ohm load.

It's a decent amp built and sold at a particular price point.
 
#19
Generalizations are useless stuff from an engineering standpoint- apologies to the more specific replies. Changing your semis is a fools errand unless you have test equipment and enough knowledge not to have to ask the question here. Caps are another story. Though I'm a big fan of measuring them to determine what's going on, changing out the low voltage ones in particular is bound to be beneficial, even if it doesn't affect offset much. I'd look at resistor values and tolerances in the front end, but the design may not do much better than you see. It wouldn't be crazy to design an offset control that injects a bit of current or changes the balance so you can adjust it to zero. Just because the factory didn't include it doesn't mean you can't DIY it.
 
Last edited:
#20
The input transistors SA1016 are excellent, to my ears smooth and neutral. SA970 sounds harsh/hard in comparison, but some might prefer that kind of sound. I can't see that 80 mV offset would be a problem. You could adjust the resistor values R607/608 or R605/606, but I don't think you would gain anything by doing it and risk to destroy your amp.

It's very satisying to DIY stuff and get it to work properly, and I think it can lead to a strong placebo effects. Take tweaking reports with a grain of salt.
 
Status
Not open for further replies.
Top Bottom