Sunday 29 January 2017

Electro-Harmonix Deluxe Memory Boy

Another quick one. This is the high-end version of the Memory Boy that I've looked at before. The Deluxe adds an extra gain control, tap tempo, tap divide, a rate control for modulation, an effects loop and the ability to control rate, depth, feedback or delay time with an expression pedal.
I got this used with a fault, the seller said the delay signal was very quiet compared to the dry. I plugged it in, and sure enough the delay was kind of weak. The manual for this one says that the added Gain control goes from -6 dB to +20 dB. When I put it straight up at 12 o'clock the delay seems to match the dry signal in terms of volume. I can't find anything wrong with it so I'm just taking a look inside.



Both PCBs.

The guts are similar to the Memory Boy, 4558/LM324/TL072 opamps, SA571 opamps and of course the 4 BL3208A BBDs. There is an ATMega16 microcontroller doing the tap-tempo and probably modulation waveforms as well. There is a JTAG port but I don't know how to dump the ATMega through JTAG, and I'm fairly certain it will be read-protected anyway.

BBD board, front
BBD board, backside

The BBDs are on a daughterboard, as per the Boy. This time they are connected with header pins and sockets and held down with screws into brass inserts instead of soldered to the main board. On the backside is the same 4011 IC used as an oscillator for the BBD clocks.


Main board uncovered

Underneath the BBD board we can see an LM13700, most likely modulating the clock frequency in the same way that the Memory Boy worked. I was a little surprised as I thought the ATMega would generate all the clock signals, it must just generate control voltages to the LM13700 to adjust the clock rate.

One nice feature is that the full IC part numbers are on the silkscreen, which should make life easier for anyone attempting to repair one.

The tap-tempo features are nice, but the rest of the pedal is essentially the same core delay as the Memory Boy. The tap-tempo does seem to allow longer delay times than the knob, around 1 second (by ear).

I'm not the biggest fan of this series, I prefer the Ibanez ES-2 for analog delays. This version is definitely an upgrade over the original if you can afford the real estate.

Sunday 22 January 2017

Line 6 DL4 rebuild

This is the first DL4 I ever got, bought broken almost 5 years ago. Despite working on DL4s for other people and my own MM4 repair I have never been able to fix this one. It has always been something I had lying around and would take out for an hour here and there whenever I had an idea.

Board as I had been storing it. The taped on diode was a part I had replaced.

In retrospect this was a bad example to try to start with. Most of these 4x4 pedals have power supply issues or have a bad IC somewhere and can be debugged without too much hassle. With this one, I replaced the some parts in the power supply back in 2012 and got the right voltages, but it would never start up. Very occasionally the LEDs would flash on and then it would freeze or die again. Looking at the test points didn't help, it would show a master clock but no other clock signals. The MCU would usually reset correctly on power up, but would then assert the reset line of the DSP and never release it. The only good model for what was going on was that multiple ICs had some kind of damage, maybe from a severe power supply failure.

Somewhere in halfway through the process of replacing all ICs.
I had tried swapping some parts with the spares PCB I had (which is now almost completely depopulated) without any luck. I'm not clear on the exact timeline of work as things were quite spread out, but I know that after replacing an MCU on one of the other DL4s I bought a box of new ICs and replaced the one on this pedal, which didn't help. At some point I decided to order all new ICs, and I replaced everything except the DSP (DSPB56364AF100) and the Cirrus audio codec (both end-of-life and hard to find new) - no change. After swapping a DSP56k part in the Whammy pedal I built up some confidence and decided to see if I could get a replacement part for the DL4.

DSP56364

I got one from a Chinese vendor on eBay, knowing that it could be a fake part. It turned up a month or so later and I put it aside. I dug out this pedal recently and tried swapping it out. The old part came off easily but I had some trouble aligning the replacement QFP, in the end I just cheated by drowning the pins in tacky flux and reflowing the board with hot air and it soldered perfectly.

It was still dead, and now the 3.3V supply was being pulled low. It looked like I had a solder bridge somewhere on the DSP, or it was a remarked/fake/dead part with a short from 3.3V to ground. My method for finding shorts to ground on a board has always been to guess where the short might be and remove parts until it's gone. I had read about a technique for finding shorts where a constant current is fed into the power rails and a DMM on millivolt range is used to measure the voltage drop across parts that connect to power and ground. The part with the lowest voltage drop is the short - this is basically a way to make low resistance measurements with a higher sensitivity than most ohmmeters can manage. I fed in 400mA to the 3.3V rail and measured all the 3.3V coupling caps. Voltage drop was lowest near the DSP (shit) but of the 4 power decoupling caps around the DSP, C39 had the lowest drop. I removed C39 and the pedal powered up, current consumption was back to normal. I don't know if C39 was bad (it didn't measure as a short outside of the circuit) or if it had a small solder bridge I couldn't see, but I was pretty pleased with this and will use this method in future.
Coolaudio parts: V1000 & V4220M. Later removed the adapter boards.

At this point the 4 LEDs would just flash on and off, but this isn't too surprising as I had removed the audio codec and the DRAM chips. I used the modulation EEPROM that came with my MM4, as I knew that program doesn't need DRAM. The Cirrus CS4223 codec is harder to source, but I happened to have some Coolaudio V4220M chips that I bought to try with their V1000 reverb chip. This looks like a second source/clone of the Cirrus chip, so I installed one. This gave me a working pedal! I tried the DL4 EEPROM that came with this pedal and it was completely dead. I moved it to my MM4 (which has been converted to a DL4) and it didn't work there either. I reinstalled the original DRAM IC and tried the MM4's known-good delay EEPROM and only got heavily distorted noise. This seems to confirm that all of the ICs that came with this DL4 are damaged.

I now have an MM4 that acts as a DL4, and a DL4 that only works as an MM4. With Behringer/Coolaudio parts in it.

I have ordered some DRAM chips that I think will work, so hopefully this can also be used as a DL4 again.

Alive.

The parts were not too expensive (€20-30) but I sunk a lot of time into this. It's satisfying as a personal victory after having this thing around for so long but if I got one in this state as a repair job I would probably turn it down.

Lessons learned:
- a complete digital rebuild is totally possible
- it's probably not worth the cost of parts and time
- all program information is stored in the EEPROM, no ICs need to be programmed or configured
- DSPB56364AF100 are available from eBay seller "e-best_trade". Unfortunately, none are listed right now
- Coolaudio V4220M works as a replacement for Cirrus CS4223
- short finding with a constant current source and sensitive voltmeter can work really well