Author Topic: Vintage Digital Guitar Processor: Digitech GSP-5  (Read 4048 times)

0 Members and 1 Guest are viewing this topic.

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Vintage Digital Guitar Processor: Digitech GSP-5
« on: September 17, 2016, 10:30:41 pm »
I have been sitting on this for some time now and I decided it was time to either fix it, or get rid of it. Note that this repair is more for my experience/learning  than it is to make it profitable. Although I am not willing to spend a whole lot of money, I wouldn't mind taking it further than you would usually.

The GSP-5 is quite a complex beast and I cannot say that I have all parts of the schematic (attached) figured out.

I have replaced the DAC (U39) with a new one and that fixed a distorted sound (when none was desired ;-) that was the initial problem.

The problem is currently that it generates a tone without any input. I have tried to switch around the DRAM the DSP uses but only when I remove them all, is the tone gone. When I insert even one, the problem comes back. I was hoping to isolate a bad DRAM chip.

When switching the unit on, after the CPU/DSP has booted up, there is a clear tone 'visible' on the output. After one second or so, the scope image becomes a jumble of lines. I thought this might be oscillation in the analog circuit but to be honest, I have no idea how that works...

Control of the unit, recalling presets and responding to buttons, driving display etc. all works fine.

So anyone willing to take a look at the schematics and help figuring out what could be wrong...

Here are some notes for the schematic (I always do this to help me understand the schematic):
Analog circuit:
LF353N Dual Opamps
LF311N Voltage comparitor
LF357N Dual Opamps

Digital circuit:
74HC157 Quad 2-input multiplexers
LH2464-12 (DRAM) (4464 in schematic)
VGT7764-6024 (DSP)
LH5116-12 (RAM)
HD63A03RP (CPU)
HY6264LP-12 (Static RAM)
PCM54HP (DAC)   - warms up - replaced it.
« Last Edit: September 20, 2016, 07:44:20 pm by obiwanjacobi »
 

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #1 on: September 20, 2016, 07:45:13 pm »
I guess this is the wrong crowd...

Offline zucca

  • Supporter
  • ****
  • Posts: 1812
  • Country: it
  • EE meid in Itali
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #2 on: September 21, 2016, 12:14:56 am »
Something has killed the DAC, I guess a bad transient on the Supply? Maybe other guys are gone on the same power rail?
I am not 100% sure but do you have a scope? Are you able to trace back that tone? Is it present at the DAC output? If yes look at then the problem must be in the digital circuit part (I am sure you have already understood that according to your description).

If nobody is giving you an answer don´t give up but contunue to post you investigations... I promise you someone will jump in and help you.

BTW can you upload some pictures? Board and Scope screenshots?

Don´t give up.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #3 on: September 21, 2016, 02:55:28 am »
Thanks for answering.

Something has killed the DAC, I guess a bad transient on the Supply?
Maybe other guys are gone on the same power rail?
I don't think it is the supply. All digital ICs are 5V and like I said the control of the unit still functions. There's a separate CPU and ROM for that. I have replaced the DAC - its brand new. Of course a bad supply could ruin that...

I am not 100% sure but do you have a scope? Are you able to trace back that tone?
Yes I have an old analog 20MHz scope. The 'tone' is output by the DAC. The analog circuitry that follows seems ok - at least good enough.

Is it present at the DAC output? If yes look at then the problem must be in the digital circuit part (I am sure you have already understood that according to your description).
Yes, the signal is at the DAC output. Indeed that is why I turned my attention towards the digital part. The DSP especially. The DSP uses 5x 64kx4bits of DRAM memory  (Schematic p1) - my guess is for sample manipulation - and have fiddled with the DRAM chips.

BTW can you upload some pictures?
Board and Scope screenshots?

On the left is the analog part and the DAC. In the center is the DSP with below it its DRAM. On the right is the CPU and ROM and power supply. The rest of it is mainly 74 logic and some static RAM.
« Last Edit: September 21, 2016, 03:00:04 am by obiwanjacobi »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 1762
  • Country: hr
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #4 on: September 21, 2016, 03:52:43 am »
Check 373 latches between DSP and DAC for stuck bits..

One of memory modules might have stuck bits too.. That would present itself as false signal...

Please explain in more detail what is a problem:

1. Does it work normally , but has tone on top of signal..
2. Frequency of tone might tell you where in sample memory it can come from...
 

Offline DaJMasta

  • Super Contributor
  • ***
  • Posts: 1416
  • Country: us
    • medpants.com
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #5 on: September 21, 2016, 05:59:44 am »
I'd definitely check the frequency of the tone, because it could tell you a lot about where it's coming from - if it's a harmonic of a signal or mains frequency then it could be cap problems or damaged shielding or something, etc.

Does the tone change at all with the input?  And as suggested, using a scope on various points on the output could tell you something about it too.
 

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #6 on: September 21, 2016, 08:22:16 pm »
I have played around some more and the 'tone' changes with each preset. A preset is a combination of settings for the different (DSP) effects. Something in those settings is used as a basis for this fault.



Check 373 latches between DSP and DAC for stuck bits..
I have checked all pins of U10 and U11 (74HC374 - Schematic p1) and all data lines are 'moving' - lots of bit patterns are transferred. There is a nice steady clk to both (OE is grounded).

One of memory modules might have stuck bits too.. That would present itself as false signal...
I have already swapped the DRAM ICs. I have swapped the DSP static RAM ICs too (U7 and U8 Schematic p1) - no change.
But I can see that on both chips
    pin 23 - A8 is stuck hi, (directly wired to DSP)
    pin 22 - A9 is stuck lo, (directly wired to DSP)
    pin 21 - /WE is stuck hi, (directly wired to DSP)
    pin 20 - /OE is stuck lo,
    pin 19 - A10 is stuck lo, (directly wired to DSP)
    (pin 18 - /CE is supposed to be lo)
    pin 16 - D6 has a consistent pulse (could be ok - could be part of mangled data)

I followed /OE back to the invertor U12F which also had a steady hi on its input, which is wired to the DSP.
So the DSP is outputting constant signals for /OE and /WE (as well).

I have pulled the DSP and cleaned its pins with alcohol, but that didn't help.

Please explain in more detail what is a problem:
1. Does it work normally , but has tone on top of signal..
2. Frequency of tone might tell you where in sample memory it can come from...
1. Yes, it appears to be. I can hear a faint guitar signal through that 'tone'.
2. See video. Time set at .1 ms / div. A few kHz by the sound of it.

Video:
https://1drv.ms/v/s!AhOnTNbY2NaPiNsSZBlKpGPNdejJog

« Last Edit: September 22, 2016, 12:00:29 am by obiwanjacobi »
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 1762
  • Country: hr
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #7 on: September 21, 2016, 11:38:25 pm »
Did those signals go hi z with dsp out. ? So we can rule out PCB.
 

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #8 on: September 22, 2016, 12:02:54 am »
Did those signals go hi z with dsp out. ? So we can rule out PCB.

You mean those stuck pins on the static RAMs?
I haven't tested. Will do that later.

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #9 on: September 28, 2016, 11:50:31 pm »
[I had an entire answer typed in and then when opening my laptop screen, the browser re-loaded and everything is gone  >:( So this is a short version.]

    pin 23 - A8 is stuck hi, (directly wired to DSP)
    pin 22 - A9 is stuck lo, (directly wired to DSP)
    pin 21 - /WE is stuck hi, (directly wired to DSP)
    pin 20 - /OE is stuck lo,
    pin 19 - A10 is stuck lo, (directly wired to DSP)
    (pin 18 - /CE is supposed to be lo)
    pin 16 - D6 has a consistent pulse (could be ok - could be part of mangled data)

A8, A9 and A10 are not wired to the DSP!, but come from U13 (Schematic p1). U13 is driven by latch U18 (schematic p2) called PD (Port D). These lines are used as sort of a memory page selector by the CPU. The other SRAM address lines are accessed through latch U17 (Schematic p2) and the 16 bit data lines with latches U16 and U30 (Schematic p2), all controlled by the CPU.

When switching presets these latches are clocked. Some presets seems to continuously activate these latches, probably to write expression pedal data (Midi) into the DSP memory (not investigated). When I lift these latches, the tone is gone. However it may simply be that the SRAM data is zero'ed out and that this will simply not generate a tone. Reinserting the latches brought back the problem (of course).

One thing I don't understand is the XENEX signal that seems to give the CPU access to the SRAM. However, I can find no way how the CPU manipulates this signal. It is connected to the DSP ENEX pin 81 (schematic p1) though...

If I had to guess I would put my money on the DRAMs and replace them all. They get the warmest -not hot exactly, you can still touch them, but warmer than I would like. I also noticed after having the unit on for a while (like an hour or so), that the little transformer gets warm and the voltage regulators (+5/+12/-12) also get warm. So perhaps something is drawing more than it should?

Thanx for reading.

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #10 on: November 26, 2016, 10:32:20 pm »
So finally the DRAM memory I ordered came in this week and I have tried it today.
No that is not it. Same tone on the output.

After staring at the schematic for a couple of minutes it dawned on me that perhaps it was just sampling garbage on the inputs...?

I started to look for the ADC but couldn't find any. So how do you convert analog to digital with only one DAC? It seems that the DAC is used to generate an value to compare against the incoming analog signal.

On page 3 the input signal is filtered through some opamps and ends up as signal 'AN_OUT' (and DRY). On Page 4 that AN_OUT signal is ultimately compared with the DAC output with U47 resulting in an output 'D'. Have not figured out where that goes yet. But this arrangement requires the analog output to be muted during this time when the DAC attempts to compare to the analog input.

And at the bottom of page 4 there is a MUTE signal but it does not move. MUTE is generated by the processor (page 2) directly. The other option would be the 'CURVE_IN' signal (bottom page 4) but that doesn't move either. CURVE_IN seems to be generated on page 7 by PD7 and PD7 originates from U18 on page 2 - its one of the ports. U18 gets a nice CLK and the input signal move all over the place - still the output is stuck. I have also seen this in other buffers U19 and U20. Would all these buffers be bad? I have exchanged U18 with U11 but that didn't change anything. I feel like I'm running in circles.

It was all the time I had so we'll have to investigate further next time.

« Last Edit: November 27, 2016, 04:03:08 am by obiwanjacobi »
 

Offline The_Penguin

  • Regular Contributor
  • *
  • Posts: 103
  • Country: ca
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #11 on: November 29, 2016, 03:15:22 am »
In case you haven't seen them, there are a couple of youtube videos on repairing these units, both implicate C29 and C30 Tantalum cap failures. May not be related, but worth a quick check.
 

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #12 on: November 29, 2016, 03:26:54 am »
Thanks. I just found them myself, haven't watched them yet - they're on my watch list.

Offline obiwanjacobi

  • Frequent Contributor
  • **
  • Posts: 882
  • Country: nl
  • What's this yippee-yayoh pin you talk about!?
Re: Vintage Digital Guitar Processor: Digitech GSP-5
« Reply #13 on: April 19, 2017, 08:02:29 pm »
After a period of rest, I am at it again...  :-/O   :-BROKE



So I have tried new parts for U40 (LF357) and U47 (LM311) that did not help. I thought that perhaps the detection of matching the digital signal to the analog signal was screwed - it appears to be okay.

I then lifted U42 (LF353) that feeds the analog part into the comparitor (U47) and made sure pin 3 of U47 was on ground. I could see a small ripple on the analog signal and thought that that may cause some problems. Although the output did clean up a little bit, the main bulk of the faulty signal remained.

There are also three pots in the circuit but I have not dared to turn them yet. I am unsure of what they do exactly (although I have some idea) and how to get them in the correct position again - I know I can mark them, but that is not very accurate.

Everything I can think of so far, still points to the digital part of the circuit, but I have no ideas how to pin-point the cause...


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf