• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Reviving a Nakamichi TA-3A

It seems as if it is working from the VAS to the output stage but lets do some sanity
checks. Q305 and Q306 form a current source their purpose is to pull a constant
current at the collector of Q305. The current is simply the voltage you measure across
R 313 put into I = V/R. That determines the emitter current of Q305, since the base
currents are low we can and are ignoring current into the base of Q306.
Now, we also know the collector current of Q305 since it is a high beta transistor
Ie ~= Ic so it is just the Ie that you calculated.
Let's calculate the drop on the 4.7K resistor, from V - IR where I is the current source
value that we just calculated. Take that voltage and subtract it from the positive supply
and that should be the voltage on the collector of Q304. We are ignoring any current
into the bases of Q307 and 308 because base currents are small for high beta transistors.

You can substitute a transistor but it must have a Vce rating of about 110V or the rail to rail
voltage. Also, compute the power dissipated at idle as Icurrent source times the rail (single)
voltage. Ic minimally would be twice the current source, but 10X is probably safer in
overload. Often I like to see a 500 mA part there but when the output is highly buffered as
it is here you can get away with less.

I'd want to be sure that nothing is broken around the diff pair so that we can be sure the
new transistor will not blow. Any electrolytic caps around the diff pair, input and feedback
to ground should be tested and really, at this age, replaced. If you put a resistor equal in
value to R303 from the base of Q302 to ground and lift one end of R314 both diff pair
transistors will have the same DC path resistance to gound and they should be balanced.
Compute the current through R307, by measuring the voltage across it. Then compute
it through R304, half of the previous value should pass through R304 if the diff pair is
balanced and working properly. Q304 should still be removed for this test.
The balance will depend on the matching of the diff pair transistors, 20% is close enough.
 
Thanks, Pete, great ... I was going to go through your suggested measurements, but found I had a bunch of a suggested equivalent 2N5401 in my parts drawers which are 150V devices 600ma current so I figured I could afford to sacrifice one in the meantime as a quick test. I had to cross over the legs on the 2N5401 due to a different pinout, but after soldering it in, the right channel came back and is working!

I'll replace it with a 2SA1145 once they arrive. Or maybe the 2N5401 is an upgrade?!

So, in summary, Q304 was the culprit of the bad right channel.
 
As per CN12 mentioned in my initial post where I asked what it is: this is the connector for the volume control LED.
 
jjbunn said:
Thanks, Pete, great ... I was going to go through your suggested measurements, but found I had a bunch of a suggested equivalent 2N5401 in my parts drawers which are 150V devices 600ma current so I figured I could afford to sacrifice one in the meantime as a quick test. I had to cross over the legs on the 2N5401 due to a different pinout, but after soldering it in, the right channel came back and is working!

I'll replace it with a 2SA1145 once they arrive. Or maybe the 2N5401 is an upgrade?!

So, in summary, Q304 was the culprit of the bad right channel.
Click to expand...

That is a very old type but it will do the job. I'd rather see the original but it is
probably fine.
 
Yes, it is certainly in the signal path but notice that the 2SA1145 is only a 50 mA part,
this is probably why it blew. It's also a 200 MHz part with a typical Cob of only 2.5pF.
That's impressive.
It's hard to say how much difference it will make but good that there is feedback
around the part.
The 2SA1145 is obsolete, how did you find one, are you sure it is not a fake?
 
Note that there is a Y suffix version of the 2N5401Y that is selected for a min
hfe of 120 and it is available at both Digikey and Arrow. That would be better
if you decide to stick with it. The thing about this part is that because it is a
higher current part it works better at a higher bias operating point, say 10 mA
or more. Try to compute the current source current to the VAS and then we
can get a sanity check.
The really good parts are vanishing fast.
 
Pete B said:
Yes, it is certainly in the signal path but notice that the 2SA1145 is only a 50 mA part,
this is probably why it blew. It's also a 200 MHz part with a typical Cob of only 2.5pF.
That's impressive.
It's hard to say how much difference it will make but good that there is feedback
around the part.
The 2SA1145 is obsolete, how did you find one, are you sure it is not a fake?
Click to expand...

Ebay ... I bought five pairs of 2SA1145/2SC2705. Seller has 99.9% +ve feedback, but we shall see :-)

Pete B said:
Note that there is a Y suffix version of the 2N5401Y that is selected for a min
hfe of 120 and it is available at both Digikey and Arrow. That would be better
if you decide to stick with it. The thing about this part is that because it is a
higher current part it works better at a higher bias operating point, say 10 mA
or more. Try to compute the current source current to the VAS and then we
can get a sanity check.
The really good parts are vanishing fast.
Click to expand...

I think I have half a dozen 2N5401 - don't think they are Y variety, but I could measure their hfe.

For current source to the VAS, is that the current into the base of Q304, or have I misunderstood?
 
Yes, thanks Pete! Very informative :-) And thanks to @northpaw for your help. I'm not sure yet whether this TA-3 is fully functional, so I may be back with more questions :-)
 
I would be very cautious about parts from ebay, there is a HUGE counterfeit
problem. It is being done by organized crime in China, India, and probably
other places. It is not just a few people. I consulted for a company with a lab
for analysis, Xray, decap, etc.
At least check them for Vce breakdown before using them.
I'd just go with the Fairchild parts, if their dissipation is more than 300mW put
a small heat sink on them to, hopefully, avoid future failures.

I helped a guy debug an Adcom GFA555 on DIYaudio and he thought it would
be best to replace a bunch of parts while debugging it and he ended up with a
few counterfeits that did NOT work.
 
Pete B said:
I would be very cautious about parts from ebay, there is a HUGE counterfeit
problem. It is being done by organized crime in China, India, and probably
other places. It is not just a few people. I consulted for a company with a lab
for analysis, Xray, decap, etc.
At least check them for Vce breakdown before using them.
I'd just go with the Fairchild parts, if their dissipation is more than 300mW put
a small heat sink on them to, hopefully, avoid future failures.

I helped a guy debug an Adcom GFA555 on DIYaudio and he thought it would
be best to replace a bunch of parts while debugging it and he ended up with a
few counterfeits that did NOT work.
Click to expand...

To test for Vce breakdown ... is that simply a matter of putting 100V across the Collector and Emitter and waiting for it to blow?!

I shall report back on my tests once the suspect devices have arrived :-)
 
If you don't have a high voltage power supply use the rail to rail voltage
of the amp. Hook up C and E through a 100K current limiting resistor and
measure the breakdown voltage across the transistor, there should be no
drop (or very minimal) across the resistor. The base connection should
be open. This assumes that your DMM has a 10Meg or more input Z.
 
Hello, i own the Nakamichi Receiver1 and it is Microprozessor controlled. After the change of the Backup Battery the unit wont´t power up, not by remote, not by anything and the power LED shows red.
After a reset of the MC-Prozessor - described in the SM- the unit powers up and all works good.

Good luck.
 
Pete B said:
If you don't have a high voltage power supply use the rail to rail voltage
of the amp. Hook up C and E through a 100K current limiting resistor and
measure the breakdown voltage across the transistor, there should be no
drop (or very minimal) across the resistor. The base connection should
be open. This assumes that your DMM has a 10Meg or more input Z.
Click to expand...

Time to dust off my 7CT1N transistor curve tracer, if I can remember how it works, but I seem to recall it can make this measurement.
 
Pete B said:
If you don't have a high voltage power supply use the rail to rail voltage
of the amp. Hook up C and E through a 100K current limiting resistor and
measure the breakdown voltage across the transistor, there should be no
drop (or very minimal) across the resistor. The base connection should
be open. This assumes that your DMM has a 10Meg or more input Z.
Click to expand...

The five pairs of 2SA1145/2SC2705 arrived, so I tested all five of the 1145 types. With a 100V supply, +ve to Emitter, Collector to one side of a 100kOhm resistor, other side of the resistor to -ve supply, I measured the drop across the resistor as less than 1mV for all five.

I then looked at all five on my curve tracer, and saw nice sets of curves for each one, nothing odd looking (although I am very rusty when it comes to curve tracers!).

Using my multimeter's transistor tester, I measured the following hfe values: 194, 164, 176, 168, 157. The spread in these numbers surprised me: perhaps it indicates some issue with quality control?
 
Controlling hfe is difficult due to process variations, it is kind of like baking
cookies - do they all come out the same?
Transistors are made on wafers and those from the same wafer should
be very similar. Your 157, 164, and 168 might be from the same
wafer. There is no need for complementary matching in the VAS location
since the VAS and current source have different functions.
Numbers seem good so far, if you have a capacitance meter and could
measure C-B value is a major clue. Cheaper, common, transistors will
have higher Cob.
 
Last edited:
Pete B said:
Controlling hfe is difficult due to process variations, it is kind of like baking
cookies - do they all come out the same?
Transistors are made on wafers and those from the same wafer should
be very similar. Your 157, 164, and 168 might be from the same
wafer. There is no need for complementary matching in the VAS location
since the VAS and current source have different functions.
Numbers seem good so far, if you have a capacitance meter and could
measure C-B value is a major clue. Cheaper, common, transistors will
have higher Cob.
Click to expand...

Thanks, Pete. I measured the C-B capacitance on all five, and they were all between 5-6pF. I guess that's a little higher than the spec.?
 
It's hard to read that low, was there any lead capacitance involved?
Yes, it is hard to read, but at least it is not 10 pF or more.
 
You must log in or register to reply here.
Back
Top Bottom