Electro-Harmonix Memory Boy #2

On to the second Memory Boy. I was hoping this would be the same problem again, but this one fought me.

After opening it, it was apparent that someone else had been in here before. The ribbon cable to the footswitch had been replaced with discrete wires and there was a large splash of solder hanging off an electrolytic cap on the main board. I have no idea if the ribbon actually failed or if the last person to work on this was just trying things, they tend to be reliable.

This one also had no wet signal. Power was OK, I checked the CLK testpoint and it also had no clock waveform. I replaced the 4011 hoping I was done, but still nothing. At this point I thought about replacing the LM13700 as well, but decided to trace out the daughterboard to see what it was actually doing. I remembered to take pictures of both sides of the PCBs this time.

Main PCB, backside

BBD board

BBD board, backside

It's not unusual when doing this that the problems reveals itself halfway through tracing out the circuit. In this case I found that I had no continuity between pin 2 of the header that connects the boards and pin 1 of the LM13700 (bias input), despite a track on the PCB clearly running between the two. I can only guess that the daughterboard was rocked on the header strip and this trace cracked. I ran a jumper wire between the header and the OTA and I had a clock signal again. Still no wet sound, which was worrying as it suggested the BBDs were bad. Fortunately, it seemed that the last repairperson also played with the bias trimpots, and after re-biasing each stage I had a working delay. Unfortunately, no schematics for everyone else, at least not yet.

Jumper wire from U12 to connector pin 2

This one sounds identical to the other, so unless I decide to try out some modifications I'll sell one (or both).

Electro-Harmonix Memory Boy #1

This was an eBay purchase that I bought as "For parts or not working". According to my account history I had been sitting on this for years before digging into it (yikes). I actually have two of these, the second unit was bought a year later and will be added here soon.

This is a 30 - 550ms analog delay (BBD) with modulation. Modulation waveform is switchable from triangle wave to square wave and rate can be set slow ("chorus") or fast ("vibrato"). An expression pedal can control delay time or modulation rate. The comparison to their legacy line is pretty obvious - this is meant to be a cut-down version of the Deluxe Memory Man, it has the same delay time but is missing controls for modulation rate and output level. I would guess that the Memory Boy and simpler Memory Toy were introduced to compete with MXR's successful Carbon Copy.

The pedal passes a clean signal but has no delay effect. Turn the blend control to 100% wet and it's completely silent. So input and output buffering and mixing are probably fine and the problem lies with the BBD circuit.

This is fairly heavily populated on the inside. This is the typical modern EHX style with pots, switches, footswitch, jacks and DC connector all mounted to a large PCB and then anchored to the case. A spring grounds the board to the enclosure. This style was designed to reduce assembly costs but it's hard to say if the quality is effected. It's a pain to get these boards out of the enclosure, I have been using a screwdriver to push in the DC jack but they get marred, so a plastic or wooden tool would work better. I like that the 3PDT footswitches is on it's own board with a ribbon cable, this switch will see the most force so strain-relieving it from the rest of the main board is a good idea.

It's a two PCB assembly, all surface mount ICs and passives. The main board appears to handle the input/output buffering, switching, modulation LFO and controls. The smaller daughterboard holds the BBD chips (4x BL3208A), the clock circuit and biasing and balancing trimmers.

Everything is analog, and apart from the BBDs EHX are only using common jellybean parts - SA571 companders, TL074, LM324 & 4558 opamps, LM13700 OTAs. These pedals should be repairable as long the SMD packages can be reworked (they can). The daughterboard is soldered to the main PCB with a 6 pin header strip, and there are some rubber pedal "feet" underneath to support it. I'm not a huge fan of this type of construction, anchoring the small PCB at one edge will allow it to lever and possible damage it. From what I've seen, the Memory Toy uses a similar construction method but the daughterboard is supported at two sides and is socketed instead of soldered. I desoldered the daughterboard to see if there were any backside components (none).

Main board - EC-D54 Rev. B

BBD Board -EC-D54 Rev. B. U13 was removed at this point.

After checking voltages at several ICs I started poking around with the oscilloscope. CMOS logic is used alongside analog ICs so 9V can be used everywhere, there are no voltage regulators onboard.
There is a helpful "CLK" test point on the smaller PCB that showed that I was getting no clock signal to the BBDs. There is a HEF4011 quad NAND-gate in use as an oscillator and an LM13700 configured as either a voltage-controlled resistor or capacitor to vary the frequency of oscillation, controlled by the modulation LFO.

I couldn't see any activity on the NAND chip (U13), so that was suspect. Lifting off the chip showed how it was connected on the PCB. I replaced the 4011 with a new one and the clock signal returned at the CLK test point. The clock frequency varies from about 5 kHz to 100 kHz. For 4 2048 stage BBDs this gives delay times from about ~40 ms to ~800ms, which seems correct. This means that bandwidth is limited to a few kHz at longer delay times,  which gives the characteristic dark sound. LFO is around 0.5 Hz in the slow ("chorus") setting.

After putting it back together it sounds great. I had some pretty bad clock noise at maximum delay time but this went away when I got it back inside the enclosure. Longer repeats are dark, and I don't hear much distortion (though I'm not sure if I would recognize if it was there). Modulation is nice, I find the triangle wave much more useful than the square. It doesn't do the clean repeats with strong attack that digital delays allow, but that's probably not what it's meant to do either.

I don't know why the 4011 failed. I did notice that the specific part used is only rated for ~1mA output current, it's possible that driving the input capacitance of the BBDs at high clock frequencies exceeded this (?) To be safe I used a MC14011 that can supply 8.8mA.

As for scope for mods, a dedicated modulation speed control would be nice. This can be done through the expression pedal jack but as it's only a single resistor in the LFO circuit controlling the rate I may attempt it there. It doesn't really oscillate when feedback is at maximum, so a series resistor with the feedback pot might have be jumpered to get it to break into oscillation.