Alesis SR16 - irritating noise problems

[Circuitbenders]

I was just trying to bend an Alesis SR16 and it was going surprisingly well. Theres loads od good stuff on the twp ROM chips but, and its a big but, i'm trying to wire up a patchbay and the cabling is picking up what sounds like loads of digital noise from the rest of the circuit or possibly radio frequencies from elsewhere.

If i solder a lead to many of the pins on the ROM chips it will pick up noise depending on where i position it. The longer the lead the more the noise and waving it near the back of the circuitboard around the amp areas picks up some strange whizzing and blipping sounds. Thsi is from one lead, after 24 of them the noise floor is just unusable

Can anyone suggest a way to get rid of this noise? Would placing a ferrite ring around the patchbay cabling help in any respect or am i thinking about something entirely different there?

[nochtanseenspecht]

maybe with shielded wire ,conect to ground on one side of the shield...but you know..
it helped with my guitar. but a lot of work for a hole patchbay..
or coppertape between wires and noise source, connected to ground.
don't know if it helps against digital noise ..

[catweazle]

seems to be an EMV problem, a ferrite core could help.
Better use single RF Choke (Ferrite Bead) for each wire (in series) as near at the IC as possible.

http://www.maplin.co.uk/search.aspx?MenuNo=12452&MenuName=Chokes&worldid=3&FromMenu=y&doy=search

Value of the Choke should be uncritical (1uH..50uH)

good luck

[catweazle]

Other things you can try:
Add small capacitors in the range 10pF to 1nF to the signalpin at the IC and the other capacitor pin connected to ground (near at the IC).

Add a resistor (around 120Ohm) in series (forms a lowpass filter with the capacitor)

Shielded cable is an option.

To be absolutely shure add an resistor at the other end of the cable to ground (termination impedance/resistance to prevent cable reflexions).

Difficult to say whats the best solution, it is case dependent (what kind of EMV problem it is).
   
PS:
be careful with your capacitor values, if they are to big you bend the circuit unwanted
(i use caps in that manner (signal --||--- ground) in my CASIO CT380 for bending with famous results)

[Circuitbenders]

hmmm, thanks for the advice. I've tried a ferrite core with absolutely no effect so i think i'll try an RF choke on the wires picking up the worst noise and see if that helps.

I'm thinking that i may end up having to use shielded cable, which is going to be a nightmare job terminating all 24 cables to the same ground. The same with the capacitor/resistor solution but if nothing else works that it'll just have to be done 

[catweazle]

Maybe you have some luck ...
Try the following:
Insert a RF choke or a 47 Ohm (minimum 1/2 Watt, not the small ones)  into the powersupply +  and  - .
This helps in case that the ground is digitally messed (fast signal changes produce differentials on the ground potential).

Each logic IC needs blocking capacitors on the supply (+)  pins to ground (near as possible), maybe the manufacturer has left them away to save some money (build as cheap as possible, it works but is not 100% stable)
Add a 100nF cap on the Vcc pin on the ROM (see datasheet),
in most cases it is the pin on the top of the right side. The blocking cap absorbs the peaks resulted by fast signal changes in the IC.

Try that thing with a resistor on the end of the cable (1k or so) first.
You build a lot of antennas on the data/adress bus which receive a lot of humm and radio frequency.
That terminator resistors avoid the reception of such signals.

sorry that I can't give you a explicit and universal solution for this problem. High frequency and electromagnetic compatibility (EMV / EMC) are mysterious worlds

[catweazle]

But wait a minute....

is the ROM in the Alesis SR16 a old 25xxx type ??
If so then all the ideas with caps, terminators, shielded cabels help nothing!!!

The problem is well known (i have asked an old school electronics engeneer) for old (EP)ROMs 25xxx-Series.
The ROM inputs are very very senistive for disturbance (no schmitt trigger inputs like on new ones).

Solutions:
- read out the ROM and burn it into a new one (the newer ones are stable),
  but you need a eprom reader/programmer for that 
or
- connect a TTL busdriver 74HC244 or 74HC245 on the data/adress lines (as near as possible on the ic)
  every TTL logic IC will do the job.

Hope that helps a bit more...
   

[Circuitbenders]

Not had a chance to try anything else out yet, mainly down to buying a load of parts and then cleverly leaving them at my girlfriends house (DOH!) but i've taken a look at the SR-16 IC's and they are a pair of TC534000P's.

I've tried to find a datasheet but although i can find a spec sheet that describes them as 'General Purpose Mask Programmable ROM' i can't actually find a datasheet with a pin out diagram anywhere.

Anyone know the location of a datasheet?

[catweazle]

Ok it seems to be "modern" CMOS technologie
Try to put 10k resistor (less or more, try...) 
from EPROM pin to ground for a stable I/O level

the TC534000 is a  512k x 8 mask rom  and can be replaced by a  27x040 EPROM
or 27c4001 EPROM. Or you can put a (non volatile)RAM instead of the original EPROM
sort of a sampler/drumcomputer ( BQ4015 NVRAM from TexasInstruments for example)


Datasheet for replacement part (27C4001)
http://www.st.com/stonline/products/literature/ds/2385/m27c4001.pdf

Here is a pinout (standard pinout on most 512kx8 RAM or ROM devices)
Interesting for bending would be the Axx pins, i think.

http://www.euro-arcade.de/files/romref.txt

Code Select Expand

[tt]
(4M) 512k x 8 mask rom                    JEDEC 27C040 EPROM
     TC534000,
     MB834000,
     23C4001) <pin 1,22 prog CS,CS/ pin 24 prog OE,OE/ on 23C4001>

nc   1 +-v-+ 32  Vdd                    VPP  1 +-v-+ 32  Vdd
A16  2 |   | 31  A18                    A16  2 |   | 31  A18
A15  3 |   | 30  A17                    A15  3 |   | 30  A17
A12  4 |   | 29  A14                    A12  4 |   | 29  A14
A7   5 |   | 28  A13                    A7   5 |   | 28  A13
A6   6 |   | 27  A8                     A6   6 |   | 27  A8
A5   7 |   | 26  A9                     A5   7 |   | 26  A9
A4   8 |   | 25  A11                    A4   8 |   | 25  A11
A3   9 |   | 24  OE/                    A3   9 |   | 24  OE/
A2  10 |   | 23  A10                    A2  10 |   | 23  A10
A1  11 |   | 22  CE/                    A1  11 |   | 22  CE/,PGM/
A0  12 |   | 21  D7                     A0  12 |   | 21  D7
D0  13 |   | 20  D6                     D0  13 |   | 20  D6
D1  14 |   | 19  D5                     D1  14 |   | 19  D5
D2  15 |   | 18  D4                     D2  15 |   | 18  D4
GND 16 +---+ 17  D3                     GND 16 +---+ 17  D3[/tt]

[Circuitbenders]

OK, well this is all a bit weird, i tried some of the suggested methods and none of them seemed to really help much but then i got hold of another SR16 cheap and it turns out that the second one i got hold of doesn't suffer from the same noise problems. Or it does, but to a far lesser extent so while i might have a try a bit of noise reduction using the suggested methods, its probably usable as it is.

This is a bit odd as the first machine i worked on is labelled as REV: F with the date on the OS Eprom saying 2003. The second machine that doesn't have noise problems appears to have exactly the same circuit and board but its labelled at REV: C with no date on it but i can only assume its a lot earlier.

It seems strange to me that the earlier revision seems to be a lot less prone to noise problems, unless of course i damaged something on the later machine somehow 

[Circuitbenders]

Ok, problem more or less solved. After trying RF chokes and various resister layouts 've managed to get the noise down to managable levels with the simplest method i.e. using shielded cable for the volume knob connections (i had to move the volume knob to mount the patchbay) and rather than using individual wires for the pacthbay cabling i've replaced all the internal wiring with a single shielded 25 way multicore cable.

Its a bit of a pain to work with as the cable is so thick and getting the case closed again was a bit tricky but i think i've got there in the end

[Circuitbenders]

[andy_wheels]

\mr wheels crawls out from under his rock.

I am always battling noise on bent things. can be right annoying... this thread has loads of good info. nice.

nice compact patch bay on that thing...

any sample sounds we can hear?

[Circuitbenders]

I'll start anotherthread once i get get some demos sorted.

It turns out that its the PCM54HP chip on the SR16 that appears to be actually producing the RF noise. Using the first couple of wiring methods i tried the individual wires were all passing right over this chip and picking up load of noise from it.

If you keep all the wiring as far away from this chip as you can and if you have to run wires near it, keep them shielded and perpendicular to it instead of running them along the chip, then you get a lot less noise.

The PCM54HP is actually the 16 bit D/A convertor so it seems a bit strange to me that its producing so much noise. Any ideas why this would be happening catweazle?

[catweazle]

I think wires over/near the PCM54 DAC produce some crosstalk (noise).
http://en.wikipedia.org/wiki/Crosstalk

---------------------------------------------------------------------------------------------------
From the datasheet:
http://focus.ti.com/lit/ds/symlink/pcm54.pdf
INSTALLATION CONSIDERATIONS
If the optional external MSB error circuitry is used (PCM54),
a potentiometer with adequate resolution and a TCR of
100ppm/°C or less is required. Also, extra care must be
taken to insure that no leakage path (either AC or DC) exists
to pin 27 (PCM54). If circuit is not used, pin 1 (PCM54)
should be terminated to common with a 0.01mF capacitor.
The PCM converter and the wiring to its connectors should
be located to provide the optimum isolation from sources of
RFI and EMI.  The important consideration in the elimination
of RF radiation or pickup is loop area; therefore, signal
leads and their return conductors should be kept close
together.This reduces the external magnetic field along with
any radiation. Also, if a signal lead and its return conductor
are wired close together, they represent a small flux-capture
cross section for any external field. This reduces radiation
pickup in the circuit."
----------------------------------------------------------------------------------------------

A big and good grounded metal housing/plate over the DAC section (PCM54 and signal traces to the PCM)
may help (shielding with coppertape for example) . I don't put my fingers into fire that it works but it's a common technique to reduce crosstalk and EMI probs. shielding shielding shielding ...

Digital signals have very fast changes between their states (0=0V, 1=~5V).
These fast potential change (0->5V; 5V->0V) produce some overshoot.
http://en.wikipedia.org/wiki/Overshoot
This produces higher frequencies (some sort of a little HF transmitter) which interact with
the DAC (or other signal traces). So long wires act like antennas sending and receiving a lot of garbage from
outside and interact between each other.

PS:
cool bent! that switchbox is cool.