Component C29 is backwards on all TB-303's... (more inside)

[Computer Controlled]

OK, is there a difference between 1uF and 1.0uF? Common sense tells me no. But i just wanna make sure.

[Brassteacher]

OK, is there a difference between 1uF and 1.0uF? Common sense tells me no. But i just wanna make sure.

You are correct.

What you may be hearing when the sound gets ugly for a few seconds then calms down is the cap building up enough charge so that it doesn't leak current severely.

[Computer Controlled]

When i first reversed it sounded just fine. It wasn't until i reset it back and put the resistors on it that it started. I think i heated the cap up too much and basically ruined it.

[Computer Controlled]

This is what it sounds like now...

http://tweakcollective.com/mp3/new/comp ... x0xb0x.wav

[aminoacid]

that sounds nasty.

was this recording from a cold and new started x0x?

is the schratchy noise apperaing when you turn resonance knob or any other knob?


the warbleing sounds like that your cap whent bad or just lost a god soldering connection after desoldering and resoldering it.


replace it and try again. have it reversed id say... it makes a nice difference. although i think the feedback is pretty cool too.

if you have the cap as the schematic say and add the resistor will you still have ugly peaks on the accents when tha volume is full and the resonance is zero?

[Robin-Whittle]

This is a 50 volt capacitor (at least in the machines I have here)
running reverse biased at 5.333. It wouldn't be surprising if there
was little or no leakage current through such a capacitor.

I tested two. One had no detectable leakage at 5.333 volts - less
than 10 nano-amps. The other had 0.85uA leakage. So it was
behaving like a presumably ~1uF capacitor with a 6.7 meg ohm
resistor across it. This would not upset the circuit in any audible
way.

I have never encountered any problem with this part of the TB-303
- and I have worked on 300 to 400 of them.

I doubt the capacitor would lose its capacitance. At worst, there
would be growing leakage current due to holes in the aluminium
oxide layer, which are not being self-healed by the electrolyte,
because the polarity is the opposite of what it should be.

I have no reason to think that electrolytic or other capacitors from
the early 1980s have anything wrong with them. I don't recall any
failing of their own accord in TB-303s.

I guess a capacitor could leak so much that its voltage would drop
in a manner, such as to 3V or so, where the voltage across R106
(2.2k) is reduced sufficiently that the current through the Q21
pair is reduced so much as to cause clipping.

I think it would have to be a drastic drop to affect the gain of the
Q21 differential amplifier to any noticeable degree. If the
leakage was noisy, or erratic, that would probably cause audible
problems.

I think it is quite likely that in all the TB-303s made, few, if any,
of these capacitors have ever failed to the point of audible
changes. Nonetheless, I will replace them in the future, with a
similar capacitor connected the right way.

In the first revision of the boards, there is another reversed
capacitor, again due to a PCB layout slip-up. This is C51, the
2.2uF 50V cap on the input of the headphone amplifier. The first
revision boards had the dot and so the negative terminal on the
R159 pin, while it should be to the IC14 pin 2 pin. Again, the
reverse voltage is small compared to the 50 volt rating of the
capacitor, and it would take a lot of leakage to affect the
behaviour of the headphone amp. This capacitor only has
voltage across it when headphones are plugged in, so for many
TB-303s, it has only rarely and briefly reverse biased.

I can't imagine a good reason for reversing the polarity of C29
on a x0xb0x.

- Robin http://www.firstpr.com.au/rwi/dfish/

[antto]

Hi Robin, and welcome

[Tim Stinchcombe]

Hi Robin,

I guess a capacitor could leak so much that its voltage would drop
in a manner, such as to 3V or so, where the voltage across R106
(2.2k) is reduced sufficiently that the current through the Q21
pair is reduced so much as to cause clipping.

I think it would have to be a drastic drop to affect the gain of the
Q21 differential amplifier to any noticeable degree. If the
leakage was noisy, or erratic, that would probably cause audible
problems.

Some of the OP's measurements as reported here show that reversing the cap can cause such drastic drops, their affect indeed being that the gain of the diff amp is sufficiently reduced so as to be audible, as several of the audio samples in the original and subsequent posts show. My post here shows how the gain will change with any leakage present by 'reversing' the capacitor, simply emulated (via SPICE simulation) by adding a resistor in parallel to C29, and sweeping its value. Further simulation results showing the affect on the overall frequency response of adding a 30k in parallel to C29 are shown in this post - to my mind these illustrate what is likely a quite tangible difference, and probably accounts for the audible differences in the audio samples.

I think it is quite likely that in all the TB-303s made, few, if any,
of these capacitors have ever failed to the point of audible
changes.

On looking over the thread whilst preparing this response, I believe it is clear that some 1uF capacitors when reversed at C29 are capable of producing audible changes - what is not so clear to me now is whether the caps that were tried backwards actually include the original cap that the TB-303 being used for the measurements was built with, or whether they were all a random selection from the OP's bag of pulled 1uF caps - perhaps he can elucidate?! (And apologies to anyone else who has reversed their cap and reported a difference - I haven't re-read the entire thread, only what I see as the larger 'key' posts..).

Tim

[antto]

i connected a 220K resistor in parallel with C29, the result was: less resonance, and the overall gain of the filter dropped a bit
then tryied with a 100K resistor - even less resonance, and less gain
i compensated the resonance by boosting the resonance boost mod (R97) and it sounds normal (tho, gain is still less)
note: my C29 is still backwards

[spdk]

Why did this conversation stop on Tue Jan 11, 2011 2:42 ?

Although I don't understand a lot of what is written, it seems to me a very important discovery.

[antto]

i don't know but, IF i ever get a x0x kit to try and build myself - i'd put LOTs of sockets in MANY places

[gurth]

I got a new x0xb0x with new components in it, I put c29 backwards and put tm3 fully open, to get some kind of resonance.

first I really was surprised and kind of liked the sound, the overall volume with and without accent was better, the clicks with filter closed weren't annoying anymore it had a more dirty sound I had a huge difference between saw and sinus The only thing was the resonance was not there anymore.....

It kind of looked like it was a good thing to do, but it was too much, If I could bring the changes down to 50 percent of what they are now I think It would sound awsome.


Does anybody know what Roland put on c29? Can I get them somewhere new or second hand?


Thanks very much

[antto]

yes, when i flipped C29 the filter output gain got lower, and the resonance became weaker
but i have the Resonance boost mod (R97 replaced with 10K pot) from before, so when i compensated for the resonance loss - it sounded pretty normal

i couldn't really see any other difference

iirc, the 303 uses Philips capacitors

[cleaninglady]

I was looking at the TB303 Service Manual and this section looks very confusing.

If i was building these in a factory from this diagram , i'd be confused.

Correct me if i'm wrong here but it looks like Diode symbols representing Capacitors here.

Also the actual C29 symbol is probably the most blurry and indeterminate one of the lot in this area.

Doe's anyone think this may have been the reason for the C29 reversal in the Roland Factory ?

C29.jpg (181.73 KiB) Viewed 4485 times

[Tim Stinchcombe]

Correct me if i'm wrong here but it looks like Diode symbols representing Capacitors here.

To me it looks like they are that cap symbol with the parallel plates with a little 'hatching' between them, as sometimes used to designate electrolytic caps (the quality of that section of the manual is possibly the result of several generations of photocopying, then the scanning, i.e. it is likely not representative of a 'factory fresh' PCB!?).

Doe's anyone think this may have been the reason for the C29 reversal in the Roland Factory ?

From memory (read: I'm not going to check back the rest of the thread!), it is the actual polarity marking of the silkscreen which is wrong, rather than the cap being placed incorrectly against the silkscreen (i.e. they were being built correctly according to what the silkscreen says).

Tim