MF10 Switched Cap filter problem

[gez]

Does anyone have any experience with MF10 Switched Cap filter chips and if so have you ever managed to get quiet operation from them?

I’ve breadboarded the Penfold Phaser and the background noise is intolerable.  It sounds like a TV/Radio that isn’t tuned in to a station.  I’ve paid strict attention to layout but it’s as noisy as hell.

RC filtering of mains doesn’t help and besides, the problem seems to be coming from the clock.

I have a number of ideas for these chips but need to get past this problem (so I don’t want recommendations for other phasers thanks, I just breadboarded the Penfold circuit as it’s a ‘proven’ design and seemed a good starting point).  Any ideas, or are they always this noisy?!

The schematic is on-line somewhere but I can’t find it so apologies for not posting a link (anyone help?)

[J. Luja]

I built that project a couple years ago and had identical results. I tried pre/de-emphasis and companding and still couldn't make it useable.
I think I tried a higher supply voltage and it made some difference, I could be wrong though.

hope you have better luck

-Jeremy

[Paul Perry (Frostwave)]

Is the 'noise' present when there is no input?

[gez]

Thanks for the replies lads.  The noise is nothing to do with the audio part of the circuit, it seems to be clock feedthrough.  This got me to thinking that the clock level is too large so I scanned through the data sheet again and, sure enough, the level shift pin "should be biased at +5V for CMOS clock levels in 10V single-supply applications", i.e. half the supply.  

The Penfold circuit has this pin tied to ground!  If I've got time this afternoon I'll try this out and report back.

[Paul Perry (Frostwave)]

I honestly think that all Penfold circuits should be approached with some trepidation. I know some work, but there have been too many wiht errors, compared to most other published collections.

[gez]

Hmmm, the other connection doesn't change anything.  Neither does running the clock at a lower voltage.  

I have another schematic for a phaser using this chip and it doesn't even have a connection for the level shift pin!  :roll:  This schematic parallels the sections.  Perhaps this was done as any feedthrough just gets amplified by the next stage if cascaded.

The datasheet (why don't I read these bloody things before I buy!) states that typical feedthrough is 10mV with a Q of 10!  :shock:  The Q is nowhere near that high, but it does suggest that they're noisy.  Perhaps they expect you to actively filter after the circuit?  

So much for the low parts count if that's the case...

[Paul Perry (Frostwave)]

Some comments on the MF10 here http://www2.charlielamm.com/synth/#mf10
but, what is better, plenty of other DIY guitasr & synth stuff as well.

[gez]

Parts wise, once you add in the buffering, bypass caps, resistor networks, yadah yadday, there are a lot of parts on your breadboard. Although it seems simple and sexy, the MF10 really isn't as simple as a basic OTA or vactrol based VCF.

Once I started to read the data sheet and factor in all the additional circuitry (not to mention the current consumption of these beasts) I started to think along the same lines (time to dust off my stash of OTAs?).

Still, I'll give it another go tomorrow.  The clock is probably too low, but I can't see anyway round this as the filter has a centre frequency 100 times lower so, with the values he's suggesting, the starting frequency for a sweep will be around 1kH...a little high for a guitarist (and probably why his filters sounded weak).  Buffering the clock might be worth trying.

Thanks for the link though Paul, anything appreciated!  Perhaps they ought to be renamed Switched Crap Filters?

[Paul Perry (Frostwave)]

There isn't raeally anything wrong with switched cap filters as such, but for audio apps, at the low end of the audio range, the X100 is getting close to audio, and maybe there are aliasing problems.
if it sounds like martians are playing along with you, that's aliasing...

[gez]

if it sounds like martians are playing along with you, that's aliasing...

I see martians playing along with me but don't hear any...that's another matter though.  No, there's no aliasing problems Paul.  It sounds OK in that respect, though this phaser is nothing to write home about (quite weak for two notches AND I've messed about with the Q to get a stonger effect!).

I think the problem might be with the 4046 as they're not that consistent from circuit to circuit and, if I recall, they vary with supply.  I'm going to get rid of the flip-flop,  Penfold is using it to divide by 4 to get the lower notch, but this (theroetically) brings the clock down to 20kH (I have a feeling that it's lower due to the 4046 being slightly out and that this might be the problem).  I'm going to try clocking the 4046 from 40kH (one schematic I've got does this) and see if this gets rid of the noise.  Failing that I'll take it higher (though it won't be much use to me if this is what it takes to cure the problem!).

Bank Holiday today so I'll have the whole afternoon to try and get this sorted out.  I'll let you know the outcome.

Thanks for your input Paul.

[gez]

Thump...did I just hear a towel being thrown in?!...certainly sounds like it!!!  :cry: / :x / :oops: / :evil: / :roll:  etc

[puretube]

i.m.h.o., the MF10 has never been intended to use as an audio filter,
but to filter out clock-noise in hybrid circuits like BBD...
i.i.r.c., the app.notes call for passive filtering after the MF10;

[gez]

i.m.h.o., the MF10 has never been intended to use as an audio filter

I have a couple of schematics where it's used for audio and both sound poor.  I had a nice idea that would have used a couple of MF10s but the filters I concocted (noise aside) didn't exactly inspire me so even if I had managed to get rid of the noise it probably wouldn't have sounded that good.  Oh well, I'll earmark the idea for a later date and do it the 'long way round'.  

the app.notes call for passive filtering after the MF10

Can't remember if the app notes do, but I included simple RC filters on input and output and it didn't make much difference - these buggers are noisy!

It was my own fault really, I should have done some more reading before plunging in...but they looked so tasty in the catalogue!