2 More DL4s

Big Green Delays, again.

I had a couple of these Line 6 DL4s handed over for repairs and each had a different problem from what I have seen before, so I'm holding out that this will be interesting to someone.

Silver-Taped Green Delay

By now I have something of a checklist of problems for these pedals. This one had all power supplies present, and none had any strange ripple or noise superimposed. Next step is to look at the test pads - there was no clock signal anywhere. Since the MCU generates the clock, and the MCU is responsible for resetting and waking the DSP (which produces the sampling clocks) nothing will work without this master clock.

U7. Oscillator and caps swapped at this point.

I know the problem is likely to be in the area of the MCU (U7). I replaced the oscillator crystal and capacitors with parts from the scrap board to see if I got a clock signal, as that's an easy place to start. No luck. I took off the MCU with hot air and replaced it with the one from the scrapper and everything worked again. Fantastic.

After removal and wicking of U7.

Alignment of the replacement is not great, but solder joints look good and function perfectly.

These MCU are a cheap 8051 core in a "ROMless" configuration, meaning that they have no ability to be programmed and only load code from external memory. This is a Good Thing for repairs, as it means that a scrap unit isn't really necessary, any new IC of this model should work, and all the code is in the one socketed EEPROM. In practice, I have found that "ROMless" is sometimes a misnomer. MCUs often have fuses or configuration bits that need to be programmed, or even have a small area of programmable memory for a bootloader, so a straight swap isn't always possible. More on this to come.

Normally I use a Kester flux pen for surface mount work, but this was the first time I used a syringe of tacky flux paste. It is really amazing stuff and I am completely won over. It made this job much faster and cleaner (I used it again for the Whammy repair, but unfortunately I'm getting to write about these out-of-order).

The owner had cover this pedal in silver gaffer tape. I think it looks cool.

A New Power Issue

This unit didn't get as far along before something look wrong. It had a working 3.3V supply, but all other power supplies voltages were low.

Another Rev 6 board.

The 6.6V voltage doubler (MAX660, U20) follows the 3.3V supply, so I started there and removed the output cap (C48) and IC. Measuring resistance across the output cap pads (6.6V to ground) gave 29 ohms, so there is still a problem.

U20 & C48 pulled.

The 5V regulator is next (U21, LM3480) so that was pulled, and the 29 ohm to ground was gone. I pulled U21 from the scrap board and replaced it, then repopulated U20 and C48. All power supplies returned and relays clicked, worked again.

Poor focus, but U21 has been replaced.

LEDs and relays are working after C48 and U20 are back.

At this point the scrap board I have is looking pretty bare. Between the 3 pedals I have written about and another personal unit that hasn't been fixed yet I've removed pretty much every IC. This is fine though, as I found out that very little (if anything) on these pedals is pre-programmed. In nearly every case new parts will work perfectly, having a donor just means avoiding waiting a day for replacements to be delivered. This shouldn't really be surprising as these are consumer goods made in large quantities, it makes sense to keep any custom, selected or programmed parts to a minimum to keep costs down.

Digitech Whammy 4

I have a bit of a backlog of stuff I've done but not found the time to describe here. It's probably best to do so before I forget everything.

Big red pitch shifter. Another 90s digital classic. I used to own one of these these but moved it along because I didn't really get on well with no polyphony and the artifacting didn't always sound too great. I think these older Whammy pedals are pieces that need to have music written with them in mind instead of just switching them on and playing the usual stuff.

A pedal large enough that the photos look terrible

Sold as not working and had had a true bypass mod done. I was hoping that it just needed calibration or that there was a simple issue with the true bypass work. Instead it just flashed all the LEDs once when powered up and then did nothing.

Inside we have a large PCB that fills most of the enclosure. There's another Motorola 56k DSP56362 (same as some of the Line 6 and EX stuff), an Atmel 89C55 (8051 architecture), a Cirrus CS4224 Codec, the MIDI optoisolator and some CMOS logic chips and opamps. The DSP56k/8051/CS4224 seems to be a really common choice of components for late 90s/2000s digital, it's very similar to what was in the Line 6 pedals I looked at previously.

The mod work was pretty bad, the wiring used was way too thick and was under strain, there were ferrite beads cracked off the PCB and the 3PDT footswitch fell apart when I was removing the old wiring.

Main PCB and LED/encoder PCB

Atmel 8051 MCU under the sticker

This has the annoying 80s/90s practice of including an AC power supply to create positive and negative voltage supplies instead of using switching converters and running off 9V DC. There are 3 linear supplies, +5V and -5V for the analog stuff and 3.3V for digital. The 3.3V was down near 0V and the regulator looked scorched, so the heatsink was desoldered and I replaced it. No change, so there is excessive current being pulled somewhere. 3.3v to GND measured around 40 ohms. I removed any 3.3V bypass capacitors in case they were shorted, and then noticed that the main DSP chip was running pretty hot.

At this point I gave up for a little while. The DSP is a 144 pin QFP and in it can be difficult to find reliable suppliers of these 56k series parts these days. There are also no schematics for this pedal online, so I would have to trace things out or make guesses.

I tried contacting Digitech and asking about a schematic, and they replied with a PDF within an hour! Here it is https://drive.google.com/file/d/0ByVCt2OFhXnyZS1JZFlmb0Z3VTA/view?usp=sharing&resourcekey=0-vBilBFlfU0YChMUrFsavrg

I also found that this IC is actually still stocked at a lot of distributors, just not in huge numbers. I decided to get one from Farnell for around 18 euros, Digikey have them as well.

Removed the original bad DSP was difficult, as the chip is quite big and there is a large top-side ground place surrounding it that takes a lot of energy to heat up. After not much progress with hot air I flooded all the pins with some Chip-Quik low-temperature solder that I had. This kept everything melted and worked well but made a mess, there were blobs of solder all over that region of the board that had be wicked and scraped off. I realised afterwards that I should have tried using a larger air nozzle for a chip this big.

After DSP removal, there is solder everywhere

After wicking and cleaning the board

 The new chip went on easily enough with some tacky flux. The true bypass mod removes the original momentary footswitch, but this switch is still needed to calibrate the footpedal, so it is usually replaced with a pushbutton mounted to the back of the enclosure. In this case, the switch was connected to the board using the original connector and then fixed down with a big blob of hot glue (you can see this in one of the pictures above). Unfortunately, all the rework heated the board enough that this glue melted and ran all over the board, leaving a thin layer around the MCU. Now nothing happened when powering it. I could peel off  some of the larger pieces of glue but most of it needed to be scraped. At this point I gave up temporarily again.

DSP resoldered,Glue cleaned, C52 replaced, Y1 removed, cleaned and repopulated

I came back a couple of weeks later and used some IPA and cotton swabs to get off the rest of the hot glue. C52, Y1 (crystal oscillator), the momentary switch connector and some of the legs of the MCU were the worst effected, so I removed all of those except the MCU and cleaned the board as best as I could. Y1 was pretty badly coated in glue, so I cleaned it and repopulated. I tried poking around wit an oscilloscope probe and found that every time I touched one of the pads of C52 the LEDs would flash. The schematic says that this is the reset timing cap, so presumably the MCU was never being reset. I replaced it with a 1uF film cap I had on hand and the Whammy fired up. I can only guess that swapping the DSP was successful, but that the hot glue was preventing Y1 from oscillating and the MCU from getting a clock signal, or C52 was bad, or both. After redoing the true bypass wiring everything works.

There is also a 1k serial EEPROM on board, but it is connected to the DSP and not the MCU, so I think it for saving calibration settings and not the program code. I would like to dump the MCU if possible, but I'm going to leave this alone. If I come across a scrap board I'll pull the chip and try to dump it. There is also a JTAG debug port wired to the DSP, I haven't looked at that at all.

Was this worth it? Replacement parts ate up a good portion of what I would have gained from selling this, and I have more than a few hours invested in it, so probably not. The newer models add polyphony and have lowered the value on these older units too. I'll add a LED (mysteriously missing? Maybe someone killed this halfway through a mod) and sell it on. I'm glad to have swapped out a QFP144 successfully and found a schematic for this, so I'm still thinking of it as a win.