4046 Oscillators progressively getting worse over time

[LektroiD]

I was running these modules at 12V, and just had another idea to increase to 15V, although it made no difference, I thought I'd mention it.

Can someone explain to me the sync line that I need to ground...

[diablojoy]

well you could try a 100K resistor to ground on the sync in jack

[diablojoy]

hmm actually there should be one on the board already that does that job
have to check the schematic.
edit: ok there actually isn't
wrongre-edit : try a 100K sync jack side of R31 to ground just kludge it on top or just on the sync in jack itself , shouldn't hurt anything and may just solve your issue ... very maybe edit yet again :wrong ah thomas you clever man
ignore this bit it will effect the sync circuit pulse length perhaps in a bad way
re-re-re-edit : because i overthink everything
just ground your sync jack input to test out this theory first. edit :then see my next post for a more certain way.

[diablojoy]

Had a bit more of a think about this on the drive home from work
To prove conclusively if the fault is in the sync circuitry or not
Remove the 4046 from its socket and very carefully bend pin 5 out so that when the chip is reinserted into the socket pin 5 will not connect through
then attach a short lead to ground to pin 5 with a small alligator clip or solder it temporarily.
fire the module back up and see if the VCO is stable now or not
if the VCO is now stable we can go from there [ not too much to check] if it still not stable at least we now know its not the sync circuitry.

[LektroiD]

I soldered a 100K resistor between the two wired pins on the sync socket. It made no difference. I'll try lifting & grounding pin 5 on the 4046 next and let you know if that sorts anything

[fonik]

it still gets me thinking: this cannot be a common failure due to a cold solder joint, or a single bad cap in the sync circuitry. these would be singularities and would not cause 4 VCOs to get bad over time...
it has to be an error that applies to all four builds.
so it has to be something that you used or did for all four VCOs (parts?). this thread has become so long overt time so i cannot remember: did you use the same parts from the same batches for all VCOs?

[fonik]

[LektroiD]

[Paradigm X]

You can actually get two free samples of the MAT12 matched pair from analog devices if you apply, thats what im using.

edit: link

https://shoppingcart.analog.com/ShoppingCartWeb/AddModel.aspx?locale=en&sc=true&ACTION=SAMPLE&modelNbr=MAT12AHZ

Another option could possibly help? eliminate a possibility?

Mines finished but need to wire up the pots etc..

Good luck with it.

[diablojoy]

LektroiD
bit of a long shot but thought i should ask anyway . which way did you mount your matched pair of 2n3904's ?
are they facing each other or same direction - are your pinouts correct ?
not sure if this has been covered before
just read back through the main topic , page 11 Argitoth had an issue,
his first post could be construed to be maybe similar.

[LektroiD]

If the image is not that clear, the thermal paste is between the flat sides of the transistors...

I'm now wondering if the thermal paste on the tempco has become conductive...

---

RICH3307.jpg
Description:
Filesize:	467.81 KB
Viewed:	640 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[diablojoy]

OK that looks right to me , oh well it was just a thought .
nice method with the tranny mounting by the way, very neat
but the thermal paste being electrically conductive is possible [some are, though most are not and also to varying degree's, also some are capacitive {a few pico farads}] . It should say if it is on the packaging .

[fonik]

[fonik]

BTW great idea how you mounted the transistors. never seen this before. very elegant!
i think for future PCB layouts i will keep the super-matched-pairs footprint and use yout method for assembling hand matched transistors.

[inlifeindeath]

also, there's a run of 2n3904 with a C suffix (IIRC) that have a different pinout. i had this issue a few years ago.

[Blue Hell]

Thermal paste FTW!

Of course .. that's it

Anyway, seems very likely.

[diablojoy]

if it is the thermal paste being electrically conductive, being smeared all over the smt tempco like that would basically short it out , that would definitely cause a whole bunch of variation with the frequency over time and fits the bill for the most likely cause given the symptoms.
easy to check if a bit tedious cleaning every trace of all that paste off.

for an idea once you get your 2210's
my preferred method till now has been with socketed ssm2210 with the tempco underneath inside the socket, I either melt a couple of slots in the ends of the socket for the leads or use 2 rows of machined SIL pins , means the chip is still easily removed if need be , but you do have to get the height of the tempco right very slightly above the socket height and with a little extra length on the leads which gives the tempco some spring to press against the underside of the chip. With an smt tempco as you have used this should work fairly well as there will be good surface contact and temp changes from stray cool air is minimised by the socket body and chip
thermal paste should not be needed at all in that case.

[LektroiD]

Well I de-soldered the tempco, wiped off the thermal paste, and re-soldered it. I have allowed it to run for a few hours to hopefully stabilise before making any dependable tests.

Now the wave does not jump around like crazy like it did (jerking from one frequency to another like a demented sample and hold unit). However the pitch is increasing and decreasing gradually like there's a really slow and subtle LFO controlling the pitch. So omitting the paste seems to have alleviated the main problem, but it's not cured it completely. Better, but still not usable...

[Ralphperez]

[Uncle Krunkus]

Wouldn't you get more accurate tracking by sandwiching the tempco between the transistors?
You really want the tempco to be enveloped by the temperature of both transistors. As it is, one side of the tempco is tracking one transistor, and the other side is tracking the air temperature above it. The temp of the lower transistor is probably not affecting the tempco very much at all.

I'd clean all the paste off everything, butt the tempco leads off the ends to keep it nice and slim, and put it between the transistors. Keep the position of the transistors almost the same as it is, just bend the lower one down slightly, upper one up slightly, thus making a very thin slot between them for the tempco.

[prgdeltablues]

Sounds like you have solved the main problem!

You could also make a cover for the transistors/tempco out of expanded polystyrene. How slow and subtle are the changes you're still hearing? If you've got the module out on a bench, could just be caused by drafts.

Peter

[diablojoy]

seems like uncle's answer has covered it all rather well.

[LektroiD]

Well, I got one working, after cleaning the thermal paste from the tempcos (I left the paste sandwiched between the transistors). The fluctuations looked huge on the scope, but in fact they were minuscule when I linked them up to audio.

The other 3 have developed other faults somehow in all the messing about...

Oscillator 1:
Random pitches prevail, and added to the calamity, the Pulse Width dial also acts as a pitch control (in reverse).

Oscillator 2:
Same as oscillator 1, except the pitch is more stable, same problem with the Pulse Width affecting the pitch.

Oscillator 3:
Nothing happens until you turn the pitch dial round to 3 O'Clock, then a fixed high frequency sound is present - pushing the dial further round makes no effect, neither does the fine tune. Again, the Pulse Width does control pitch, this time in the standard orientation (clockwise for higher).

Oscillator 4:
Working ok.

I'm wondering in all the fiddling around if I've possibly done something to the PW part of the circuit which is causing the rest of the circuit to malfunction. I understand this is on the smaller PCB, so I will detach that from the main board and run the tests again...

[diablojoy]

[LektroiD]