advertisement


Modifying NCC200 boards

Why not just use it and have fun? Or do you have a better preamp now...?

BTW: what about extrem star earthing in the power amp like in the Starfish...?
 
Why not just use it and have fun? Or do you have a better preamp now...?

BTW: what about extrem star earthing in the power amp like in the Starfish...?
I use a miniDSP Flex as a pre-amp these days to cross over a sub :)
 
Well after a bit of fiddling, I managed to get a meaningful reading using my 3.1kHz low distortion oscillator and filter. (I'd been suffering high levels of rectifier pick up).
Using 100R resistors to feed the bases of the drivers and with a pair of 2R5 emitter resistors on these to drive the feedback I get a very clean output dominated by noise. Loading this with 100R being driven at 4V rms, I get a saw-tooth wave of twice the fundamental at 80dB down:
20220619-201607.jpg

This shows the fundamental of 11.3V p-p and the residual of about 1.1mV p-p (filter has x10 gain, limited to about 0.25MHz) so -80dB. I need to build this into a proper output stage and drive a proper load, but suspect that it would make a very nice headphone amp, particularly as driving half the load halves the distortion so driving 600R headphones (eg Sennheiser HD600) then any distortion would be below the noise floor.
 
Here's the modified board:
20220624-224233.jpg

I currently have it set at 25mA bias to get the distortion pretty much as per the image posted previously. Much lower than this level of bias and the distortion creeps up (zero-crossing spikes).
Trouble is, after running for 1/2 hour at 4Vrms into 7R5 dummy load, this more than doubles the bias current to 60mA. This is with +/-30Vdc bench psu, so the board is burning about 16W. Transistors run at just under 50C as I can place the back of a finger on the transistor cases and keep them there so they are hot but not excessively. I can understand why the NCC220 uses thermal control. Also noticed that the transistor mounting surfaces are quite poor, even when tightened down, the fit isn't brilliant as there are tiny gaps at the sides. Any more thermal compound and I'd have run the risk of seepage between the insulator and the resin which could have shorted one or other pins to the case. (My thermal compound is electrically conductive)
Edit: in fact a bit of further measuring shows it is only running 20mA bias when cooled down and the switching spikes are about twice as big as when warmed up for a while as above. Will have to consider using a npn on the U-section.
 
Used a BD437 for thermal compensation:
20220625-163201.jpg

Tweaked the bias current with my 3.13kHz oscillator with the amp at 4Vrms output into 7R5, so 2W.
This is the distortion, same as that posted above:
20220625-163045.jpg

Very little change after 20 mins at this output. Output transistors are running about 50C as I can just stand back of my finger on the cans for 5 secs. Bias current measures near as dammit 65mA and rises to 95mA as it cools down.
Well pleased!
Edit: I changed the 2k7 bias string resistor to 3k3 to get enough bias adjustment, and swapped the 47pF Miller cap to 82pF.
I'll put some square waves into it later to see how it fares...
 
1kHz square wave, 8V p-p into 7R5 so 2W:
20220625-174913.jpg

Tiniest bit of overshoot.
10kHz, same as above:
20220625-174846.jpg

No overshoot nor ringing.
Symmetrical slew rate of 4V/micro-second.
 
Nice results.

The slew rate is probably limited by whatever the input RC filter is - you'd need to connect your signal source after this to find the limit, but then-again - that's also the very point of having an input low-pass filter - to keep the amp out of HF trouble!

And
, even so, your 4v/uS is quite 'fast' enough for a full 100w into 8ohm - i.e. more than the amp is capable of.

Do keep us updated on where you go from here : )
 
Before I changed the Miller cap from 47pF to 82pF I was getting about 7V/us slew rate at 10kHz, but with a bit more overshoot at 1kHz. Qudos uses 68pF, blameless uses 100pF. I had 82pFs in the parts box... :)
I'm going to try the 'improved turn-off strategy' from the blameless on the other board (220R//few uF) across the emitters of the drivers instead of 100R from each driver emitter to the output, just to see...
 
I tried 220R//470nF (blameless circuit uses 1uF) between the drivers having lifted the pair of 100Rs. There was maybe a very slightly cleaner residual, but the basic shape of the distortion was as shown previously, so I put the 100Rs back.
I now have one circuit built up with the split earth return as per Post 6, and have built the second board with the 'reinforced' earth as per Post 1.
I have run both with the additional circuit as follows:
20220627-154654.jpg

This allowed me to inject some ac current into the from end to simulate the amp PSUs 'sagging' under low frequency/high power output (which is asymmetrical.) Putting 0.99V rms across the 560R gives a current swing of 5mA pp and this modulates the front end 0V, giving 1.6mV pp on the output with the amp idling but reduces to 0.6mV pp under 2W load at 3.13kHz:
20220627-134705.jpg

As you can see the residual has gone up, but unfortunately I couldnt photograph it at low scan rate to show the 30Hz ripple.
Both circuits tested the same, so it is down to the impedance of the earth return wire, and much simpler to just bridge the front end earth spade connections. :)
 
One of my 220 amps popped on Friday, a second time a Kendiel cap failed in a cap 6 board, had a spare board so while I had them apart I moded the decoupling cap track as you suggested.
I have some version 4 ,200 boards and the tracs are exactly the same.
It defiantly sounds a bit better cleaner than before so nice one virtually free improvement.
Out of interest what transistors did you use on the current source and VAS mods.
 
Oh dear, that's a shame/nuisance. Do you have dc speaker protection?
I used a pair of bc556 in the current mirror to replace the 1k and 22k resistors to tr1 and tr2 respectively, each with 68R emitter degeneration resistors; and bc556 as the additional transistor in the VAS mod, along with the original ztx753. Just remember that the 'blameless' is 'upside down' compared to the NCC200/220.
Edit: Dont forget that ther are also other mods (full complimentary output stage) and I'd need to draw the whole circuit out for completeness.
 
Last edited:
Oh dear, that's a shame/nuisance. Do you have dc speaker protection?
I used a pair of bc556 in the current mirror to replace the 1k and 22k resistors to tr1 and tr2 respectively, each with 68R emitter degeneration resistors; and bc556 as the additional transistor in the VAS mod, along with the original ztx753. Just remember that the 'blameless' is 'upside down' compared to the NCC200/220.
Edit: Dont forget that ther are also other mods (full complimentary output stage) and I'd need to draw the whole circuit out for completeness.
Yea had protection onboard, last time one failed it destroyed the outlet stage literally ,this time just a pair of rectifier but it blew the tracs around them so had to replace whole board.The 220 board was fine no obvious damage, I have a couple of 200 boards spare so might try some of these ideas on them.Noticed the Blamless was upside down compared to the NCC boards did throw me for a minute.
I think Les mentioned at some point that he had added a current mirror on the 300 boards and increased the current driving the outlet stage.
 
This essentially is the front end:
20220628-195330.jpg

Then you have to cut tracks and reconnect to change to full complimentary output pair including removing the 100R between the driver emitter and the negative rail and replace with a wire link, plus remove Baxandall diode, thermal tracking (tr5 replaced by medium power NPN, in my case a bd137 on heatsink) 2k7 on bias string replaced with 3k3, 47pF Miller cap replaced by 82pF to 100pF and combos of 680+470pF//560 and 390+470pF//680 each replaced by 100R.
Usual caveats apply about this being at your own risk, mains voltages are LETHAL, etc.
I have set the bias across the pair of 0R22 resistors to 42 to 44mV, i.e. 95 to 100mA bias current.
 
I run my NCC300s with the driver emitters not connected to the output, which is supposed to improve the charge suckout behaviour of the output devices. IIUC that's what AP is doing in post #31.

DC offset protection can only save speakers, not the amp!
 
@S-Man I tried this and used 220R//470nF instead of the pair of 100R and it made beggar all difference to the residual, maybe slightly less high harmonics. Certainly didn't change the saw tooth distortion at 0V crossing to any notable degree. Running at 33R load, this switching distortion disappears into the background noise.
 
I think it will only make a difference at high current.
I have other reasons for doing this in the NCC300.
 


advertisement


Back
Top