Replacement transistors, NAIM NAP90

Are the red circled resistors values the actual ones on your amp ?.
 

Attachments

  • NAP90.jpg
    NAP90.jpg
    54.1 KB · Views: 465
Ok , replace Q3 emitter resistors (56R) by a 470R as a first step.

Assuming all the transistors are good this should reduce output DC offset
to a few dozen mV.

The 27K resistor should be reduced to 1K for faster setting of the circuit
at power on and better caracteristics as well.

BC239 are 30V devices , so the max power supply should be +-30V ,
BC550 are a better choice if ever you have some.
 
Last edited:
Member
Joined 2010
Paid Member
Re #38.
The voltages you measured tell us the 752 is turned on and the 652 is off. On the good channel both are partly on, so there is a DC balance between them, as with all the voltages measured there. Since you've replaced all the semis and the problem remains on the bad channel, there is a fault in the remaining resistors or PCB tracks or even a connection or part remaining in circuit that should not be there. The feedback path to Q2 is still connected to the output stage resistors and the limiter caps. Don't rule out soldering shorts or open connections as a cause either.

Make sure it's not the signal input as there is no input capacitor shown which would prevent DC at the input - this could have a profound effect. Like Wahab, I'm suspicious of the schematic as I think there are some parts not shown. I wouldn't mess with the intentional imbalance of the emitter resistors in the input pair transistors, though. That is what Naim sound is partly about - "fix" it and you don't have a Naim any more. The other channel is fine with them as is.
 
So basically, based on the information I provided, you are convinced there is a problem with the observed behavior, which I should sort out before placing the output stage back,
and would recommend NOT to place the output stage back as long as I don't get the readings similar to the ones on the good channel ?

Is it possible that the readings are because there is no output stage currently ,and it might become correct once it is back ?
 
Last edited:
Re #38.
The voltages you measured tell us the 752 is turned on and the 652 is off.
.

If the said resistors values are genuine then the input stage current
is as high as 10mA.

The differential will settle the ZTX752 base to 0.65V or so , hence
the current through Q1 will be only 0.65mA , all the rest , 9.35mA ,
will try to flow through Q2 via an inadequate 27K resistor .
This is likely the cause of the circuit disfunctionality offset wise.
 
I deleted every trace that is not relevant on the schematics, with the output stage transistors off...
attached here. does it gives a better picture ?

Another finding which may give us some hint (?)

If I disconnect the resistor, which goes from the negative rail to the 10uf cap,
the reading between the E and C of the BC639 (ZTX652) now become 0v,
while the reading between E and C of the BC650 (ZTX752) become 55v ...

Just the opposite then it is when the resistor is not disconnected (when E-C of 752 is 0v and E-C of 652 is 55v).

Does that give some direction ?

I couldn't see and short or bad connection. I checked it a few times. The resistor values are correct. The diodes are new and are placed in the correct orientation :/

Thanks again!
 

Attachments

  • nap90 no output stage.jpg
    nap90 no output stage.jpg
    63.6 KB · Views: 346
Last edited:
Member
Joined 2010
Paid Member
Your conclusions, #45 are about right. Earlier, Wahab raised the issue of the 56R resistor (circled) in the emitter of Q3. It looks wrong and I believe 560R is more likely. Wahab's 470R may be better by calculation from the schematic. These could be 4-band colour coded which would be green-blue-black-black or yellow-violet-black-black respectively. If 3 band instead, green-blue-brown etc. In any case, measure the voltage across the resistor and at the base (from the negative rail) of Q3. That will confirm whether the total current regulation is within reasonable bounds.

'Good idea to delete the unused parts. Now, are you referring to the 1k resistor at the base of Q2? This sets the feedback ratio. Without it, the amplifier is not an amplifier. I think the arrow pointing down is wrong though. Normally, there would be capacitor of 10-100uF from that point to ground. Is that what you also refer to? Perhaps you mean the 1.8k resistors towards the output side. These should both be removed as a measure to eliminate possibly faulty parts like those caps from the circuit. (a limiter transistor and diode there have failed already)

You will need the other channel in operation again to compare some more voltages when we get the operation more normal, with voltage levels of the signal path closer to 0V.
Exasperating, ain't it? :eek:
 
Hi Ian,

Yes, I wa referring to the 1.8K on the schematics. When the bottom one is out, the voltages flips so the reading between the E and C of the BC639 (ZTX652) is 0v while the reading between E and C of the BC650 (ZTX752) become 55v.

While tracing the circuit I found that the schematics here doesn't 100% matches the actual circuit. Values of resistors can be a little different and there is a missing part or two on the schematics.

Are you sure I don't get these odd voltages only because the output stage is taken off ?

You will need the other channel in operation again to compare some more voltages when we get the operation more normal, with voltage levels of the signal path closer to 0V.

What exactly do you mean and how do I check this ?
 
Member
Joined 2010
Paid Member
Hi
First, both 1.8k resistors need to be removed or neither - balance is required and it's best assured by removing both. It was an oversight in my request to just remove transistors. It is difficult to be specific about every step necessary to carry out a general operation like "remove output stage" for someone to whom these things are alien. Apologies, but we will be here past Christmas if we don't press on.

The quotation means just what it says; you said (#47) that the good channel was now disconnected. I simply meant that you'll need it reconnected to compare voltages again, when required. When DC voltages at the input and output nodes are near 0V again, the amplifier output stage can be reinstalled. A basic RCA amplifier design such as this one does not need an output stage for stability. All the voltage gain and feedback arrangements are already in place, so the output stage only adds the capability of practical loudspeaker current, without changing the basic DC, signal levels, gain and stability arrangements of the amplifier.

Randy Slone's book "High Power Audio Amplifier Construction Manual" (McGraw-Hill/Amazon) illustrates the procedure on a similar "Blameless" model amplifier, if you need the visual support.

The output stage also includes the current limiter transistors but it needs a bias current generator and this is situated in the VAS (ZTX 384). Whilst this only needs to regulate a bias current to flow through the output stage, its presence also affects the VAS bias current, so it needs to be there at all times, for testing and alignment of the basic amplifier, just as we are attempting.

IOW, if the amplifier doesn't work at the Voltage Amplifier Stage, it won't work in the complete form either. So get this right first by comparison of actual part values and voltages and check the schematic if you are confident to. Both seem to be in error, as you now find. We've made a few suggestions and I'd like to know what that "56R" resistor is. It should be larger than 56R to limit the current to less, not more than the VAS current, which is set in a similar manner from the same voltage reference by the 68R resistor at the emitter of the ZTX652. We know roughly what values should be there from the the NAP 140 design I linked. Without accurate knowledge of what is really there, we can only offer general advice at this distance, so keep us informed of errors or we can't really help.

If one thing is certain, it's that NAIM designs change very little between models or over time. They are a frozen design, as it were, to guarantee a certain unique sound quality.

I don't expect circuit operation to become crystal clear - It took me years of stumbling about and I had built around 30 amplifiers before I learned anything significant on my own about these DC coupled amplifiers. Don't be discouraged, as a full understanding is deep in Physics and won't come easy. Once seen in their dynamic simplicity, all is revealed but it takes a lot of groundwork to prepare for this. I certainly lack the specific disciplines of the many qualified electronics guys here. However, this never stopped me building, repairing and getting a lot of amplifiers to work. :)
 
Member
Joined 2010
Paid Member
Hi again !

After I checked the circuit again and again in order to try to find some bad connections or errors, and found nothing. So I decided there's no way to proceed other than placing the output stage in and see what I get,
hoping that the wrong readings result from the circuit being not completely installed...

So that's what I did. I am using:

BD911 for the NA005
BC550B for the ZTX384C
BC549B for the BC239C
BC639 for ZTX652 (both VAS and Output)
BC640 for ZTX752 (both VAS and Output)

A quick check (with the good channel connected) it seems I get identical readings as I got on good channel:

(VAS) ZTX752 E to C: Bad Ch: 27v , Good Ch: 27v
(VAS) ZTX652 E to C: Bad Ch: 27v , Good Ch: 26.5v
BC550 E to C: Bad Ch: 2.2v , Good Ch: 2.2v
BC550 E to Ground: Bad Ch: 1v , Good Ch: 1v
BC550 C to Ground: Bad Ch: 1v , Good Ch: 1v
Voltage between the positive and negative speaker rails: 0v

I also compared voltages at different locations the circuit between the channels,
I get identical readings, in all locations, except for Q2 collector,
which reads 13.4v on the good channel, but 16.4v on the "bad" channel...

The bias trimpot is set to the same value as on the one on the "good" channel.

Is it because I'm using different parts ?

Also, how do I bias both channels ?

Should I insist of getting genuine ZTX653/753/384/BC239 before I go on ?

Thanks !
 
Last edited:
Member
Joined 2010
Paid Member
Hi
The bias setting is monitored by measuring the current flowing through the output stage. You do this by measuring the voltage across both of the large 0.22 ohm output transistor "emitter" resistors (to get a more accurate reading) as previously mentioned. To be clear which ones, that's the same as measuring from the emitter of the top NA005 transistor to the collector of the bottom NA005 transistor. 11mV there translates to 25 mA bias - enough for a Quasi-complementary design. Take great care not to slip and short leads in doing this - we all have done it! IC clips for your meter leads make it much safer. Check the good channel, though this may not be set at that value for any number of reasons, such as someone else adjusting it.

The problem for Naim designs is the time lag. The bias regulating transistor, shown as a ZTX384, is not mounted to the heatsink, so it relies on air to conduct heat from the output stage and regulate bias. The cover has to be replaced and it takes forever to warm up, so I suggest setting it lower than recommended maximum 30mA initially, to avoid disastrous thermal runaway.

Voltage differences. Nothing is a big problem there and the Q2 voltage is likely due to different transistor type/gain.

Current differences. What is the voltage across the 68R and "56"R resistors at the bottom left of the schematic? 'Still waiting for the value of that alleged 56R resistor. We need these values to calculate the current in the input and VAS stages to sign off on whether they are OK.

I assume the bulb tester is still in place, which should not be glowing now as total DC current for both channels should be less than 100 mA.

When working, you really need to fit the original ZTX parts as discussed. You could use all BC550C for Q1 and Q2 as well and try to match Vbe within a few mV if possible. Don't handle whilst testing as even slight heat changes it.

So far-so good. :) Are the limiter transistors/diodes also reinstated?
 
Last edited:
I wonder how I can protect the amp from being burnt this way again.

I checked the input cable and what caused this disaster was the ground wire touching a channel input wire... What can be done to protect the inputs, and speaker outputs, so when either the input or speaker positive output touches the ground, nothing will happen ?

I don't care about retaining the sound quality in such cases.
 
OK this is strange - I wanted to see how stable the "new" NAP90 is. So I used the bulb limiter and let it run. After 10 minutes the bulb lit up ! I realized that both new BD911 were extremely hot. The J13009-2 on the other channel were not hot at all... I know I made a mistake by not installing the board back into the case placing the heatsink back etc. Fortunately, the circuit hasn't been damaged... Are the BD911 supposed to get that hot ?

To answer your two previous questions -

- Yes, everything is installed (including all diodes transistors)
- The voltage across the "56"R and 68 R is 0.6v (as on the good channel)
 
Member
Joined 2010
Paid Member
The output transistors should warm up slightly but that is when mounted on the heatsink. I guess it's always too late to warn when you have already done what you wanted to.

The problem is that there is no cooling and so heating of the output transistors will be rapid and run away, since there is no way to gradually warm the whole circuit and feed back enough heat info. to the bias regulator transistor to control current. It's too late to warn you about trying the circuit with no way of knowing how it is set, other than the position of the pot. That is seldom a good enough indicator and the components are different now, too.

Bias current is what heats the transistor and you need to adjust it to minimum, first thing (after properly mounting the output transistors and checking for shorts to the sink, that is).

That's what the pot. is for. Read post #54. Make safe, non shorting connections to the circuit for your meter while you test the current - it is so easy to destroy everything in a flash of carelessness with hand held probes. I suggest you turn it down to minimum or zero mV and see what gets warm then. Keep an eye on the mV and feel the output device over 10 mins or so. If it remains cold, raise it to the value I indicated and watch it again


This is delicate because, as explained, effectively there is only air controlling the bias current - it's slow and the cover is needed to get steady readings. BTW, check the driver transistors as they may have failed before those mighty output transistors!

Take it gently and you might get some Christmas music after all. :xmastree:
 
Member
Joined 2010
Paid Member
Re #57

Was this problem before we began to look at the amplifier here? Actually, simply earthing the input would not cause the amplifier channel to overheat. Quite the reverse - it would shut it down. There is a possibility that if there was a lot of hum or the connection was rapidly switching and causing a high level audio hash current to flow, you could overload it, but after all, the input has to pass through an isolating capacitor, so unless this has shorted, there is not likely to be a problem there.
http://www.diyaudio.com/forums/atta...naim-nap90-amp-schematics-naim-nap90-01ab.gif
In the linked schematic, C1 is the input cap. referred to. If you like, you can also check C3 for a short (unpowered, of course). Use the probes with the same polarity as the cap is marked.
C3 can be problem for DC on the output and if this an original small bead tantulum type, as Naims are, they can fail after long years. However, it's easy enough to test at the output terminals as a high or varying DC offset voltage.

N.B. You should also check this voltage across the output terminals when you fire up. It should be less than 40 mV and slowly falling, IIRC. In a correctly operating amplifier, it is dependent only on input transistor (Q1,Q2) matching, previously mentioned.