Kenwood KA-3500 recap

Madfro is referring to the DC offset, not the bias. The DC offset can measured at the speaker terminals +/- right and +/- left with the volume at minimum, speakers disconnected, tone controls off or set to zero. I'm assuming this is a direct coupled amp, and if so you'd look for a reading as close to 0mV for each channel. If the reading is 50mV or higher it might be worth your efforts to gain match the differential transistors in an attempt to lower the offset voltage for better performance.

Right, thanks. Brain fart on my part - sure will check offset and report back - if it's way off, I'll correct.
 
The original issue on her was that I was getting much higher DC offset on the Right channel - around 52 mV where the left was nice and low, around 4-5 mV if I remember.... I did not look at the bias current though.

Well, that's still there. Running 53mv on right channel. 14 on left.
 
Looks isn't everything.... Boosting up filtering capacitance stresses rectifiers. Engineers designed the 3500 with 6800uF
Stick with this value.
FMPOV, you won't get neither better bass rendering or whatever from boosting up filtering capacitance.
Thanks, I think you're right that changing isn' always good.
In this case, it should be OK. I just checked and I could double the current rating for cheap.
 
Madfro is referring to the DC offset, not the bias. The DC offset can measured at the speaker terminals +/- right and +/- left with the volume at minimum, speakers disconnected, tone controls off or set to zero. I'm assuming this is a direct coupled amp, and if so you'd look for a reading as close to 0mV for each channel. If the reading is 50mV or higher it might be worth your efforts to gain match the differential transistors in an attempt to lower the offset voltage for better performance.

Ok, I'm gonna get back on this and get it wrapped up.

I do want to get that offset under control. The differential transistors - these are Qe1\3 and Qe2\4, correct? These are on the "power" board.
OEM: 2SA620WN
Hmmm, KSA992 replacement, I think.

EDIT: Some confusion, or substitution - Q1-4 on board X07-1470-10 are actually 2SA640.
Replacement is still 2SA992-FBU though.
 
Last edited:
For matching transistors, it's been years for me and I just used a multimeter.
But my new cheap chinese transistor tested doesn't do Vbe. I see people building a simple circuit to make this testing easier - what do you guys do?

And found this... Hmmm.
If the impedance the Bases see is very low, hFE does not matter, but Vbe may. This is true for some diff-pairs and mirrors, the cases JCX was talking about.

If the impedance the Bases see is very high, hFE matters a lot. This can be seen in the diff-pairs of some op- and audio amps which use high value bias resistors.
 
If your Chinese meter reads Hfe that should be good enough for what youre doing. My only advice would be to try and not hold the transistor in your hand any longer than you have to. The temperature will have an effect on the reading so try to remain consistent in how you handle them with that in mind.
 
Last edited:
If your Chinese meter reads Hfe that should be good enough for what youre doing. My only advice would be to try and not hold the transistor in your hand any longer than you have to. The temperature will have an effect on the reading so try to remain consistent in how you handle them with that in mind.

Thanks and good advice about heating with your fingers. I tested and it sure makes a difference. LOL, so I set up a box fan blowing in from the side to keep things relatively even on temp.
I've got some great matches on the transistors. I ordered 50, so had plenty.

Now, I have to take that main board out again - my fault, oh well. I'm getting good at it.
 
I replaced those diff pairs with matched 992's.
Offset did come down on one channel, up slightly on the other. Both are matching around 24mv avg.
Kenwood KA-3500 17.jpg

But the offset is now fluctuating - quite a bit - about a 10mv swing on both channels.
Bias sets perfectly.
I'm tempted to find another set of diff pairs - not sure - advice?
Is this possibly oscillation, how do I test for that?
 
I also found this quote on another thread: "This was a very changing time for transistor development. The basing configuration on many of these transistors is unusual. The X-sistor diagrams are to the left of the schematic. Double and triple check the biasing when installing new transistors in the output stage." Are the new transistors in the input stage or the output stage?
 
I also found this quote on another thread: "This was a very changing time for transistor development. The basing configuration on many of these transistors is unusual. The X-sistor diagrams are to the left of the schematic. Double and triple check the biasing when installing new transistors in the output stage." Are the new transistors in the input stage or the output stage?

These are on the output stage.
I'll check that post you linked. Of course, I didn't mark the transistors that I removed, so will have to "re-match" them based on my removal notes, if they are re-installed.
And I am tempted to do that, as hopjohn suggested. But then we'll be back at the original high right ch offset, LOL.

Maybe just bringing down the offset by raising values of R15\R16 - but I'm not sure what that does to performance. A couple of people here have had the exact same symptoms, same unit, same 992 replacements, as we have here.
http://audiokarma.org/forums/index....a-3500-becomes-distorted.258824/#post-6347969

EDIT: the quote says triple check biasing (he means that biasing adjustment the manual?)
I guess I'll do the triple check.

EDIT02: maybe the quote means pinout rather than bias? It's the pinout that's to the left of the schematic... I made sure I have pinout correct.
 

Attachments

  • 2SA640_pinout.JPG
    2SA640_pinout.JPG
    17.2 KB · Views: 11
Last edited:
Last edited:
I guess that's the best bet...and try to flip around the originals to get the channels more even? It's sure better amp design when you get both bias and dc pots... sorry this is turing into such a bugger, JP!
 
OK, I'm gonna go all OCD on you guys. No conclusions from this diversion, but I'll lay it out here.
NO diversions BECAUSE I visually and physically matched Q1-4 that I replaced.
This particular post is about the service manual, all these screenshots you see are straight from the service manual.
First, note which board we're looking at, and which transistors are sitting in Q1-4.
Kenwood KA-3500_diff_pair_01.jpg Kenwood KA-3500_diff_pair_02.jpg

Now, the schematic, matching the board and transistors.
Kenwood KA-3500 transistors 04.JPG

Now, to the parts list.
Kenwood KA-3500 transistors 05.JPG

I'm sure you see the issue, installed is 2SA640 vs. the 2SA620 shown in more than one place in the schematic.

The KSA992 should replace both, so no issues.

Nothing more than a distraction -
 

Attachments

  • 2SA640_pinout.JPG
    2SA640_pinout.JPG
    17.2 KB · Views: 13
  • Kenwood KA-3500 transistors 06.JPG
    Kenwood KA-3500 transistors 06.JPG
    12.8 KB · Views: 14
Last edited:
I guess that's the best bet...and try to flip around the originals to get the channels more even? It's sure better amp design when you get both bias and dc pots... sorry this is turing into such a bugger, JP!

Meh, it's the way it goes. Of course I want it all to go in perfectly, and I knew I was having a pretty good run with those Carver units.
This one is just bringing me down to Earth a little, nothing wrong with that. I do like the work, and I am learning.
 
Back at this - during the summer, I typically ride by bike alot, and my company had gotten a contract with a really large ($$) customer that needed alot of attention. So, no work done on this during the summer.

Yesterday, I fired this amp up and the fluctuating offset is gone. I didn't do anything to the amp itself - I did rewire some of the power circuits at the bench as I had way too much stuff on that circuit. Basically, my electronics bench, the kitchen and a few other outlets including the one in which I have a little electric heater hooked up (for winter) and about 4 fluorescent . I'd blow that breaker occasionally.
So now, the electronics bench is on it's own circuit, no electric heater, no fluorescent lights, no kitchen. But really can't put things back to see if that fluctuating offset comes back.

When powering up the unit, I had also found my grounding wire had come unhooked for the amp board, which is unbolted so has no mechanical ground. Again, I'm not sure this plays into the funky offset - it's just something I learned and always do - if I break a mechanical ground, I generally run a wire to keep that path intact (our equip at work is fussy about such things).

So, my offsets were running about 27mv and 38mv. I wanted them to be close to the same - installed a pot in series with a resistor (I didn't have just the right size pot) on the left channel and just adjusted it until it was close to the right channel. Both around 27mv now
Kenwood KA3500 pot.jpg

Here's a short video of results (if you can see it - it's on G+)

https://photos.app.goo.gl/zAy6TXw8ZDL5K2rO2
 
Power is landing right where you'd expect - amp is rated at 40wpc.
Input signal at 1K hz, 1vdc - output right before clipping is 18.8 vrms.

That comes out to 44 wpc. I didn't test at other freqs.
upload_2017-11-7_13-47-48.png
 
Nothing wrong with a little toothpaste to clean the face - then I wax with a black car wax (mcguires)
Strange horizontal lines on the alum face - looks like maybe there from the factory.

Kenwood_face.jpg DSCF2568.JPG
 
UPDATE - After getting it all back together, a few days of testing showed a "crackle" in the left channel at startup, after sitting for "a while" (generally overnight). Once it's past that first startup crackle, it's OK no matter how many times it's turned on\off. I did measure some high DC at the speaker terminals when this happened, up to about 13volts.
I don't think the right channel was doing this at all.

So, back down to the bench and turning off\on via power strip showed this wasn't the power switch acting up. Thinking the 10K caps might be a bit much for the regulator, I've swapped those out for some 4800uf (remember the original caps both tested out at 2400uf) caps for testing. But I have ordered caps at the original value, 6800\50v, I'll install those when they arrive.

And I'm obsessing over the DC offset fluctuating still - 3mv or less (BOTH channels) but it just bugs me. It didn't do that (that I remember?) before the recap and diff transistor pair replacement. But I did put the original diff pair back in and fluctuation still there. I shouldn't worry about this, according to my reading on the topic, but darn it. So, I've order some 2SA970's, which someone in another thread had good luck with, and they do cross over in specs. 992's in there now, that was my first replacement choice. Those should be fine, but I'm a glutton for punishment, I guess. :)

At any rate, this isn't done.
 
Back
Top Bottom