Falcon + CUBASE AUDIO + SCSI - Experiences?

[whomper]

Prior to installing DFB1X and clock patch, DSP worked good. I had no audio crackled in Cubase, so I'm not sure it's related to power, however perhaps now that there is increased demand on the PSU, things might have changed. I did however install PSU Cap kit.

Which IC and pin # should I scope of the DSP clock?

I'll look to desolder the PSU connector. Is it C147 that should be replaced?

[exxos]

Prior to installing DFB1X and clock patch, DSP worked good. I had no audio crackled in Cubase, so I'm not sure it's related to power, however perhaps now that there is increased demand on the PSU, things might have changed. I did however install PSU Cap kit.

Have you tried the DISABLE jumper on the DFB1X , to see what happens there ?

I'll look to desolder the PSU connector. Is it C147 that should be replaced?

Assume that is a typo and should have said resolder ?

Is it C147 that should be replaced?

The big cap which is in between the PSU connector and Videl.

I would check the voltage ripple on the DFB1X 0V & 5V pads as well. You need your scope on X10, AC coupling, about 200mV/div and see whats there. Check what the actual voltage is. Needs to be 5.10V to function properly.

Which IC and pin # should I scope of the DSP clock?

dsp.PNG

dsp1.PNG

Red arrow should be the 32MHz clock. Yellow arrow might be GND..

If the clock is bad, solder something like 470R across the 2 points. It should load the oscillator and give it better output. But if it worked without the DFB1X, its probably not that anyway. But can't assume either way until tested.

[exxos]

It reminds me of another "DSP issue" in this thread https://exxosforum.co.uk/forum/viewtopi ... 56#p123056 but while it circled around the DSP, im not sure if @stephen_usher ever got to the bottom of it all ?

There is the newer firmware which you could try to see if it makes any odds, but I don't remember what changes there was other than additions off the top of my head.

I think at some point I designed a PLL clock patch to keep the EXP clock in sync with the SDMA clock thinking that might be a issue, but I gave up with the idea as my Falcon didn't care about the clock sync anyway.

You could check the clock on the DFB1X headers..

Looks like its pin 23 which probably goes to the clock patch output.

cpuclk.PNG

[stephen_usher]

I'm afraid not.

For demos etc. I disable the DFB1X.

[whomper]

Did the following:

1. Reflowed PSU board connectors and checked the large capacitor for capacity and resistance - All good.

2. Tested DSP clock, it is 32 MHz image below.

DSP Clock.png

3. Tested DFB1X 5V for ripple between board GND and +5. Actual voltage is 5.033v, image below (ripples between -140/+140mV, did I test correctly?)

DFB1X 5V Ripple.png

4. Tested DFB1X pin 23 clock, image below

DFB1X Pin 23.png

5. Disabled DFB1X using onboard jumper, SDMA errors and cradles persist. My SDMA only worked with the original clock patch I had before and not when I just upgraded the clock for V4).

[exxos]

I'm a bit confused because earlier you seem to imply the problem only happened when the DB1X was plugged in? Now your saying it worked with your old clock patch? It would help if you clearly list the combinations and results. What did and didn't work originally and what did and didn't work after fitting V4. You also mentionned thermal problems which can be confusing things more.

[whomper]

OK, let me summarize scenarios:

Scenario 1 - Stock Falcon w/ 74F08N clock patch
Cubase Audio & DMA test app - Working good. No crackles

Scenario 2 - Stock Falcon w/ V4 clock patch
Cubase Audio - Audio crackles
DMA test app - continues small red spikes (I've sent an image a few posts ago of the same behavior)

Scenario 3 - DFB1X w/ 74F08N clock patch
Cubase audio - Audio crackles
DMA test app - All good, green

Scenario 4 - DFB1X w/ V4 clock patch
Cubase audio - Audio crackles
DMA test app - All good, green

Scenario 5 - DFB1X in disabled mode, on board TOS w/ v4 clock patch
Cubase Audio - Audio crackles
DMA test app - continues small red spikes (I've sent an image a few posts ago of the same behavior)
(This is similar to scenario 2, where it behaves stock w/ V4 clock patch).

In summary, Cubase audio only worked for me in Scenario 1.

As for the heat issue, I am currently not doing anything with it and all my testing are with a relatively cool system to begin with.

Questions:
1. Also, I am getting 5.033V on DFB1X board and not 5.1V, is that an issue?
2. When DFB1X is operational, is the SDMA clocked with the FPU like when FPU is on board?
3. When DFB1X is disabled via jumper, there is no FPU connected with SDMA (I did not reinstall the FPU on board), right?
4. I have the 100pF installed as per instructions. Do I still need it with DFB1X as there is no on board FPU?
5. Since I have firmware 1.04, I can not opt to disable DFB1X when using Cubase as I need to open up the case and install TOS, not practical. Any thoughts?


Other than the Cubase audio crackles, DFB1X with both clock patches works great for me (so far).

[exxos]

I think for now , just remove the DFB1X and just troubleshoot one thing at a time.. No use trying to fix DFB1X issues when the base system is unstable to start with. You state the problem is there without DFB1X, so just remove it and make things simpler to diagnose.

I think there may be confusion on your original post viewtopic.php?t=7797 as was that with the original clock patch or the V4 ? It implies you had issues with the original clock patch, but actually you don't according to your Scenario 1 ? But something must not have been working as you started asking about SCSI issues ?

The V4 doesn't do the inversion like the original clock patch, but my Falcon and others don't seem to care either way. It is possible option if I send you a V2 board to try and see if the inversion helps or not. It would be really useful to know because there has been no evidence of if the inversion is needed or not yet.

You could try bridging over the 62R with a small wire to see if that makes any difference. But Generally while it adds more voltage (shouldn't be needed) it increases ringing more. It shouldn't make any difference but its easy to try.

The V2 removed a DC offset issue but the V4 got around that problem with better voltage drive. We can try similar .

The Falcon FPU & ROM should be removed when the DFB1X is fitted. But the cap makes no odds if the Falcon FPU is fitted or not.

If your getting "DMA test app - All good, green" with the V4 then its probably not clock patch related. Likely SCSI issue , maybe bad drive or termination. SCSI issues I can't help with as I not used it for for over 20 years. I never used Cubase on the Falcon either.

[whomper]

I've reverted to scenario 2, no DFB1X and V4 clock plus bridging over 62 Ohm without removing it.

I'm getting audio crackles with both Cubase and also FAMP mp3 player that used the DSP, plus DMA testing app has constant small red spikes.
I am not getting all green with V4. The Falcon seems to work correctly apart from SDMA.

Happy to try V2 board if you can send a copy.

SCSI termination is present on board and also in my external bluescsi, so that should be covered. I've also tested it at length copying many files around, and it looks stable.

[exxos]

I'll try and sort it out next week.

Meanwhile you can try the v2 offset fix. But you'll need to replace the 62R with 100nf 0805 capacitor. Then 100k resistor from the SDMA side to GND.

This doesn't show the capacitor but shows 100k resistor connections

DSCN3718.jpeg

Though if you're good at changing small chips on the v4 you could replace the input buffer with inverter. It probably would be a lot quicker if you can source parts locally yourself. It would do the inversion that the original clock patch does. But there's no evidence yet inversion matters anyway.. It shouldn't..but worth trying..