Kenwood KA-3500 and my blooper

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2007
Paid Member
They are not all going to be faulty. Did you do the basic tests to see if any actually read obviously faulty (short circuit) and did you measure those low value resistors ? You need to know if they have failed open circuit or not.

Just replacing devices and switching on in hope isn't going to work... you need to be much more methodical than that.

The drivers Sakis mentioned would be fine but you need to be able to source them and that could be a problem for you. So I've looked again at this again and here is what I would do.

Replace QE21 with a TIP41C
Replace QE23 with a TIP42C

Both those will need the correct insulating kits for the "T0220" type package as shown in the picture. And you have to be sure you can mount those OK on the heatsink because they are physically different to the originals. I don't envisage a problem there though.

QE17 replaced with a 2N5551
QE19 replaced with a 2N5401

The pin outs of these are different but again, that is no problem.

QE9,11,13 and 15 are part of the protection circuitry. Whether faulty or not they are no problem in getting the amp working as the amp will work correctly with them removed. If they did prove faulty then 2N5551's and 2N5401's would be fine as replacements.

If the 0.47 ohm resistors are faulty then they need replacing.

All those parts should be fairly widely available. Realistically all those transistors would be well under £5 from an authorised distributer. Plus your other bits and bobs, resistors if needed and the insulating kits. The dry type silicon washers are easier to use, the old mica type (as in the picture) need heatsink compound smearing on them to get good thermal conductivity.
 

Attachments

  • T0220 Insulator.PNG
    T0220 Insulator.PNG
    241.1 KB · Views: 201
mooly, eastern ele., et al, good morning. had to get up early to make sure the birds didn't sleep in. :)
being the admitted noob that i am, i must admit to my ignorance of (almost) all things solid state. that being said, i have a couple of questions to run by you guys.
1. can these devices be tested in situ, or must they be removed?
2. would you describe to me exactly the steps i take to test each of these items (except resistors)?
3. can i use either type of insulating kit for each of the new transistors? i do have a few pieces of mica here somewhere, and i do have the heat transfer compound. whichever of these mounting kits you think i should use is the way i will go.
4. can i mix and match both of your suggestions? i.e. use a MJE340 and MJE350 to drive, and C4467 A1694 for outputs, or, 2Sb560 2SD438 drivers and TIP41C and TIP42C for outputs. (trying to cover all bases here as the store is a good 65 kilometre drive from here, by way of 60K and 80K zones. supposed 60 and 80K zones ;) )
5. more a request for confirmation than a question. so, after testing and finding fault, and in a worst case scenario, i want/need to purchase/source at the very least:

the two .47 Ωhttp://en.wikipedia.org/wiki/Omega resistors, the two 330 Ω resistors, one MJE340 and one MJE350 for drivers, one TIP41C and one TIP42C for outputs.

OR

two .47 Ω resistors, two 330 Ω resistors, one 2Sb560 and one 2SD438 for drivers, one C4467 and one A1694 for drivers
whew. i love tubes. this stuff hurts me head. and no, the fact i used "in a worst case scenario" and "at the very least" together has not escaped me. :0
guys, thanks so so much in advance for your patience in helping me attempt a resurrection of my sons amp.
best for now,
aidan
 
Administrator
Joined 2007
Paid Member
If the transistors are short circuit (or nearly so) then that will show up in circuit. On a DVM on the diode range if you see something like 0.127 between C and E and with the meter leads both ways around then they are duff. The 0.47 ohms can probably be checked in situ, the others quite possibly not, you would have to isolate one end.

If you're relying on just one store having everything you need (and trying to cover it all in one trip) then that's tricky :D

The 0.47 ohms need to be at least 2.5 to 3 watt. But measure the old ones first. If they are OK then then they are good to go.

All the 2SA/B/C and D devices are Japanese and will be difficult (probably impossible) to source locally.

TIP41C and TIP42C are technically a matched a pair. The 2N5551 and 2N5401 are also complementary to each other. I would ring the store and ask what they have first.
 
What's up with this particular model blowing up? I see it everywhere! It is one of the rare amps using totally obsolete output transistors that truly cannot be replaced with anything.

Unlike you put in Sakis level effort. :)


Simple as that ...Marketing reasons forced many companies like Kenwood , Akai, Pioneer and even Nakamichi people ( that are supposed to be more careful) to work with the specific semis from NEC

Bottom line is that these semis played like hell, performance was astronomical for the time , semis was easy to drive, BUT safety margins and expected power was way overestimated for TO 220 size semi .

In real life very few of the amplifiers produced survived in long time run ( actually they did but failed each and every time pushed )

NEC chemicals was forced to close and/or liquidated from Fairchild ( for some period also Fairchild was selling re stamped semis originally made by NEC )
 
@Lower tones

---Measure your guts before you are ordering parts if you are up to the task of this repair in the end you will need a DVM to check bias and offset .And after that especially if you change drivers or outputs you need to scope the amp for oscillation .

---Trying to replace semis in order to troubleshoot the amp will only increase the cost and end up in a mess and a blown amplifier

---Use a DVM diode check to see which of the semis you have blown after that you make estimation on a choice of parts

---An amplifier repair is a question of method ...you start from step one and end up in step 10 ...starting from step 5 ...is obviously wrong ...

---You will get plenty of help many willing people here in the forum but you need to listen , work carefully , and follow procedure

Kind regards
Sakis
 
sakis, agreed, and thanks for pulling my reins in. i'm rearin' to go, as i'm anxious to have the thing in good order for my son when i visit him. and mooly, will do, i'll isolate the semis by either desoldering it from the board, or take the next item down the line from at least two of it's legs off the board. i have pretty well all manner of test gear here. thing is, this is the very first time i have played with any solid state gear. i have been an avid tube enthusiast for over 40 years, but this is a totally different kettle of fish. i will present you guys with my findings early tomorrow morning and, fingers crossed, you guys will have the patience to steer me straight.
i called my supplier today. no problem on the 2N5551 and 2N5401, and the TIP41C and TIP42C. will take a couple of days for them to get them if i give them the go ahead. they weren't sure on the 2SB560 and 2SD438, and the C4467 and A1694 but they will let me know by monday or tuesday the availability.
okay, so the little missus has a huge honey-do list for me this evening, but i will have a report on the state of the components which you circled on the schematic for me mooly. and with the report, a myriad of further questions.
i am so grateful for the help you both have offered to date.
talk soon,
aidan
 
Administrator
Joined 2007
Paid Member
OK :) And remember as a first step that its easiest to check for shorted devices with them still in circuit. If they are short they are short.

(And for interest I threw the basic circuit into a simulator to see what showed up and Sakis is correct on the drivers being a little problematic, the MJE340/350 combination that I initially suggested did actually show a problem that would have needed other component changes to get them to work, the 2N5401/5551's however seemed fine)
 
mornin' all. well, i have some readings that i would like you guys to analyze for me.
i removed Q17 and Q19 from the board. i left Q21 and Q23 in circuit, figuring that with Q17/19 out of the circuit, it would be a circuitous (pun!) route for the B, C, or E legs to travel, and as the DCv from my meter would have to travel through capacitors, i assumed this would be okay for the test. the caps should be enough to impede the DC from my meter. sound good?
anyway. here are the results of my measurements. with resistors, the ( ) indicates what is called for.
R29 = 388 (390)
R31 = 387 (390)
R33 = 14.7K (15K)
R35 = 14.9K (15K)
R37 = 1.78K (1.8K)
R39 = 1.81K (1.8K)
R41 = 330 (330)
R43 = 331 (330)
R45 = .46 (.47)
R47 = .47 (.47)

semiconductors, where the first letter indicates the positive lead of the meter
*Q17 reads
C-E = open
E-C = .562
E-B = .562
B-E = open
C-B = open
B-C = .535

*Q19 reads
C-E = .004
E-C = .004
E-B = .537
B-E = open
C-B = open
B-C = .535

*Q21 reads
C-E = open
E-C = .473
E-B = .232
B-E = .232
C-B = open
B-C = .432

*Q23 reads
C-E = .512
E-C = open, after an initial spike from caps
E-B = open
B-E.231
C-B = .458
B-C = open

as unfamiliar as i am with SS devices, i can certainly see the appeal of working on them. they aren't tubes, but they are fun :)
thanks again guys for your patience in helping me repair my mistake.
best for now,
aidan
 
Administrator
Joined 2007
Paid Member
Resistors first... and all those look OK.

(High value resistors never fail under fault conditions, there just isn't the voltage available to exceed their ratings)

Transistors... Q19 is dead :D

Q21 and Q23 are showing a low reading B to E but if they are in circuit they see the 330 ohms as across B and E. So could well be OK which would be great with them being an odd case style and unobtainable.

Q17 doesn't show as short thus far.

So you might get away with just replacement drivers. Don't be tempted to power this up without a bulb tester though.
 
morning mooly, and thanks so much to you and sakis for hanging in there with me.
just a couple of questions, if i may.
1. would the failure of Q19 cause the high current symptom i'm experiencing?
recall, if you would, that the cause of this amps misfortune was due to my shorting the junction of Q23 and R47 to the junction of Q19 and R43 (NOOB question alert: is this the emitter of Q23 to emitter of Q19?)
2. must drivers be matched?
3. must drivers be matched to output transistors?
4. does changing Q19 necessitate changing Q17?
5. will the drivers you suggested, 2N5551 and 2N5401, be okay with the existing 2SB618 and 2SB588 finals?
6. following your response to the above, should i implement the same changes to the other channel? would he (son) hear the difference between the channels?
7. any caveats to observe by the end user in future operations? eg. don't play at full volume, don't engage too much bass/treble/whatever, etc

i really appreciate the assistance so far. real nice spirit on these forums. been looking around at some other topics. great stuff.
ok, thanks for now,
aidan
 
Last edited:
Administrator
Joined 2007
Paid Member
Morning... it's mid afternoon here :)

Yes, Q19 being short is causing Q23 to conduct heavily. I'm surprised its survived tbh but it doesn't read short... so fingers crossed. We could go a lot further with these tests but at this stage its easiest given the results so far to just take them at face value and fingers crossed, they are OK.

Drivers should ideally be complementary types but that's a loose term tbh. All things being equal you could fit totally different and non related devices to most amps and they would still perform pretty much the same. Shhh I didn't say that :)

Do they have to be matched to the outputs... no.

If nothing else is amiss then replacing just Q19 with something suitable should work OK but ideally we normally go for complementary types, that means an NPN and its matching PNP (or vice versa).

Will the 2N's be OK. That's the big big question. As your interested I'll show you the simulation I ran. I did this just for my own curiosity :) Its basically your design less the protection circuitry. First up is the circuit (using 2N's) followed by showing the full rated output into 8 ohms. For what its worth the distortion came out at 0.006% The driver currents are surprisingly low at 45ma peak. Even with crummy 2N3055 type outputs they are still not that horrendous.

Then we have the same test but using the MJE devices I suggested earlier and which Sakis suggested could be problematic... and indeed it seems they are as you can see. Thanks Sakis :) There is oscillation and instability.

Now in the simulation (and don't read too much into this) it seems that the upper driver (the NPN) is the most critical in that fitting a much slower and lower gain device for the PNP alone didn't cause an issue.

So question 7. Given the limited (in a short time) availability of parts I think you should go with the 2N drivers and replace both. They cost peanuts. Initially though, just replace the failed one and lets see where its all at. If its OK then we can replace the upper one too. If its not OK and there are other issues then we take that one step at a time. Either that one transistor is all that is faulty or it has damaged the output but they are not showing as such given the very basic tests done so far.
 

Attachments

  • Capture.PNG
    Capture.PNG
    54.9 KB · Views: 136
  • Full output.PNG
    Full output.PNG
    54.1 KB · Views: 134
  • MJE devices.PNG
    MJE devices.PNG
    51.4 KB · Views: 120
Administrator
Joined 2007
Paid Member
There are quite a few things you can still do at this stage to test it further... if you want to. A variac makes a few things a bit more possible :)

What I'm thinking... as a test on low voltage you could use the transistors Qe9/Qe11 as a driver to see if the amp basically runs.

:)
 
mooly, wow. now that is impressive! the program you ran must save you a great deal of time and effort, not to mention parts. and, .006% distortion! is this the same as the oft touted figures used to describe total harmonic distortion used by marketing depts since way back? seems very low, does it not? was the .006 figure what you would have expected from this amp?
and BRILLIANT! swap Q9 for Q19 and Q11 for Q17. i'll have to give that one some thought, though. i was very careful when i desoldered the Q17 and Q19, fearful of damaging the board. though i have good soldering skills, i lack the experience of working with solid state gear. i can just see me lifting a trace from the board, and.... you get the picture!
i have gone ahead and order the two drivers and the two finals from my local dealer. quite the mark-up they have. about 35 pounds sterling for 8 of them (both channels).
will let you know when they arrive in the next day or three.
besides the dim bulb, variac, any other steps i should observe?
again, thank you so much for your help and encouragement in this matter. certainly a learning experience for me the whole way through.
best for now,
aidan
 
Administrator
Joined 2007
Paid Member
Yes, its quite useful program. Its not infallible though and the results (real vs simulation) depend on many factors... but yes, worth its weight in gold.

For desoldering I always recommend solder braid and to practice first on scrap bit of equipment.

Using braid,
http://www.diyaudio.com/forums/parts/127924-working-smd-how-do-without-specialised-tools.html

The important things are to be 100% certain you identify the replacement devices correctly with regard to lead polarity and device type.

For the newer replacements you can get all the data sheets from here which show the pin outs. Just type the device into the blank box and pick a well known manufacturers version.
Datasheet catalog for integrated circuits, diodes, triacs, and other semiconductors, view

Before you power it up (with the bulb/variac) we need to turn the bias on both channels to zero by turning the preset so that it has maximum resistance. In other words so that it appears as a 1k rather than a short. That will allow Qe7 to be turned fully on which gives the lowest bias current.
 

Attachments

  • Bias.PNG
    Bias.PNG
    159 KB · Views: 100
Administrator
Joined 2007
Paid Member
question: with Q21's E-B and B-E readings differ from all the other semis. in each of the others, the meter reads open in one direction, but Q21 gives a reading in both. is this alright?
aidan

You were measuring in circuit I think initially which can give misleading readings. Lets be sure... if you isolate the base and emitter you should read around 0.6 ish from B to E and the same from B to C on both devices. That is with the red meter lead of the DVM on the base for the NPN device (the top one) and the black lead on the base for the PNP. With the meter leads reversed there should be no reading.

If you imagine the junctions as diodes then this is how they should measure.
 

Attachments

  • Transistor.png
    Transistor.png
    17.6 KB · Views: 91
Ι am watching closely ....

Don't be impressed of the simulation results these are partially ideal or even wrong the models on the transistors may present different results and for the simulator transistors are perfect , do not suffer from thermal issues , power supply and ground are also clean together with absolute perfect distribution and so on and on .

In real life circuits none of the above facts exist .

Simulation is just an indication and will help you proceed if you manage to come even close to the simulation you have done a grate job !!!

Kind regards
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.