Ludwig Phase II Clone Debugging and Miscellany

[Keppy]

Possibilities from some PUT oscillators I just looked up:

1) Ralt from gate to ground instead of gate to cathode:


2) Ground the cathode by jumpering R7:


3) Keep the current configuration but adjust values. This graphic supports the configuration of the circuit as it is now, but notice R2, the extra resistance on the Gate (R1 in the graphic is Ralt in the build doc).

[Ry]

All of those are all easy enough to wire up on my build.  I'll start playing with them as soon as I can, it would be easy enough to wire up a trimpot for Ralt and tack it to ground. 

I dug the project out of a drawer and hooked it up.  It sounds far better than my original, actually.  The only thing missing is the animation, and the trimmers need to be readjusted just a little bit.

[R.G.]

One of us needs to breadboard it along with the flipflop and get it working. I have all the necessary parts, but I don't really know how PUTs work, and it'll be a little while before I can get to it, so if anyone else wants to jump in just say the word. Otherwise I'll try and get to it when I can.

I'm darn near destitute on bench time, but I'll pour it into a simulator when I can.

PUTs are an odd variant of SCR where the voltage on the gate is held fixed, and when the voltage on the anode gets bigger than the gate voltage by X amount, the "SCR" triggers and dumps the cap through the device.

[Keppy]

One more revision: The toggles for the fuzz mode and animation mode appear to be backwards, with Voice Fuzz and Fuzz Rpt. at the top toggle positions rather than the bottom. The labels should be reversed.**

**For original mojo reverse the toggles, not the labels.

[Govmnt_Lacky]

One more revision: The toggles for the fuzz mode and animation mode appear to be backwards, with Voice Fuzz and Fuzz Rpt. at the top toggle positions rather than the bottom. The labels should be reversed.**

**For original mojo reverse the toggles, not the labels.

WHAT!!??     

[R.G.]

One more revision: The toggles for the fuzz mode and animation mode appear to be backwards, with Voice Fuzz and Fuzz Rpt. at the top toggle positions rather than the bottom. The labels should be reversed.**

**For original mojo reverse the toggles, not the labels.

Hmmm... I thought I did that on the PCB in the change from V1 to V2. Did I muck that up?

[Keppy]

One more revision: The toggles for the fuzz mode and animation mode appear to be backwards, with Voice Fuzz and Fuzz Rpt. at the top toggle positions rather than the bottom. The labels should be reversed.**

**For original mojo reverse the toggles, not the labels.

Hmmm... I thought I did that on the PCB in the change from V1 to V2. Did I muck that up?

I remember Dino asked you to flip around the Fast/Slow switch, so that might be why that one doesn't match the other two. I've checked it over several times to be sure that it's inconsistent and not just a case of me interpreting it wrong. Like, I could be reading it backwards from the way you intended, but then the Fast/Slow switch is the one that's wrong instead of the other two.

[R.G.]

OK, I'll go dig through the source files and find out how I mucked it. 

Maybe by V3 I can get this thing right!   

[digi2t]

I've been digging back into this monster, and there's something that I never addressed way back when.

In the original units, there were 10 transistors that were color coded. 4 green, 4 yellow, one red, and one blue. These were TO-92 packages, whist all the rest were TO-105's. The transistors invloved are;
Green (NPN);
Q8, Q12, Q15, Q19
Yellow (PNP);
Q9, Q11, Q16, Q18
Red;
Q6
Blue;
Q5

The first eight are all found in the filter section, and the last two are part of the fuzz section. I'm wondering though, could it be that these transistors were matched, hence the color coding?
Unfortunately, I don't have either of my original units anymore to test, but if I get really ambitious, I might even breadboard the theory one day. Would running a sim be able to show whether gain matching would play a role here?

[R.G.]

Within the same type number, color codes were used for sorting. Usually that was for sorting into a few buckets of hfe. Back in the 60s and 70s, silicon fabrication was crude enough that the same process line turned out wafers that had *wide* variations. You might get one wafer with mostly hfe=50 devices, the next one through would have mostly hfe200, the next 150, and so on. The Japanese practice was to label them all the same (e.g. 2SC1815) but to color code them into black/orange/yellow/green/pomegranite/puce... bins of similar hfes. The US practice was to make several part numbers (e.g. 2N4401, 2N4403...) and hfe-test, dropping them into the right part number bin for marking. I'm sure there were other schemes as well.

If you bought plain vanilla parts from Motorola or Delco (!), you got generally hfes with a 3-1 variation inside a part number. It was possible, and many manufacturers availed themselves of this, to get the semiconductor house to more finely test the parts and provide them to you as a special or house-numbered part that was selected to your own particular needs. Sometimes these were actually numbered 80-5043-23 or something equally cryptic, and sometimes just color dotted. Or just sent as Your Special Delivery Don't Mix These Up.

So yes, there was almost certainly some kind of sorting. The problem is that we don't know and probably can't find out what they sorted FOR. We don't know if they sorted for hfe, or Vbe variance, or frequency response, or any of many other characteristics. We can guess that the transistors involved in those very differential-amplifier looking things in the filters might have needed to be matched at least for each pair, perhaps for a quad or a sextet.

We also don't know which of the pairs were actually needed matched, or for how close the selection needed to be.

I worried about this when I was working over the schematics and layout, and decided that they'd either work or not with same-type-number devices. If they didn't, I'd hear about it and could concoct some kind of matching jig, as those setups were common back in the bad old days when circuits were not well designed to ignore or minimize the effect of transistor parameter variation. I was very pleased when the units actually worked with transistors out of the bags.

That tells me that the matching can't be all that critical, and that it is likely that transistors of the same type number ordered in one batch from the same supplier will probably work fine by default. Today's semiconductor lines make BIG wafers, with large numbers of transistors cut from one wafer, so that hundreds or thousands of devices are likely to be packaged as a part number at once, and they'll all be from the same wafer and hence very similar. That was not always true. And it's possible to get unlucky and get scrapings from the bottom of the parts bin that are from several different wafers and not all that similar.

It is tough to tinker with mismatched devices in a simulator. This is one of the pitfalls of simulators if you don't know about it. Simulators use transistor models that are **identical** for every device you put into the schematic unless you explicitly tell the sim to bend them a bit. Sometimes you have to hand tinker the individual device models, some simulators have a "tolerance" or "Monte Carlo" mode that assigns random variations for you on the theory that you'll make many runs and compare them.

If you want to mess with this on a breadboard, yell and I'll try to come up with some measurement setups where you can select for and against matched devices to test. Testing HFE is easy - the Fuzz Face transistor tester at GEO does that, and for silicon you can ignore the leakage test entirely, as silicon will have negligible leakage. Delta-Vbe isn't all that hard to test and I could come up with something.

[Ry]

IT WORKS!   

I gave up on getting the 6027 running and ordered a 2646 from Small Bear.  Luckily, I had socketed the 6027, so I simply replaced the 2646 and snipped Ralt.  The animation immediately fired up and works like a charm.  It's really rough right now, rough and magical!   I need to work with the trimmers a bit. 

It sounds 100% better than the original that I have...those of you who had originals, did you notice the same thing?  I need to crack my old one open and give it a tune up, it oscillates when the foot controller is pressed all the way down.  Is there anything you guys want me to investigate while I have it taken apart?

[Keppy]

Awesome! I'm glad you finally get the full experience.

I'm not aware of any lingering questions. I finished the last of my two builds tonight, and I have a list of minor project revisions for R.G., but I thought I'd wait until we get the PUT figured out.

[Keppy]

Oh, and regarding transistor matching, the transistors in my original build are all socketed (since, you know, I had no idea if they would work ). It seems like it hasn't mattered to this point, but if anyone has a specific test with a specific goal I can get it done. Sometime.

[wavley]

I don't remember all the things that Keppy posted, but here are the things I've done to mine apart from everyone else:

1.  Increased the value of the High Z input cap to not cut as much bass when playing baritone or Bass VI guitar.
2.  Replace the input bipolar with a MOSFET as RG suggested (I haven't gotten around to this one yet)
3.  Added a volume control to the effected out
4.  Added a mixer resistor from the stereo out to the main out so that when engaged works as a clean blend (when used with the effect volume control you can make nice more subtle spaceyness)
5.  Used an expression pedal that has a wider travel that kind of goes past the filters (Jimi and I both discovered this independently)

[Keppy]

Here's the list of needed documentation updates since my earlier posts disappeared:

1) Fix the swapped inputs on the wiring diagram.
2) Add a 13-15V zener from R64/39/40 (cathode) to ground (anode).
3) Switch the labeling on the Fuzz and Animation toggles, unless you want to flip the switch positions to match the original.
4) Jumper R7 when using the 6027 to ground the cathode.
5) First page of the build doc still mentions a 5k rocker pot; should be 10k
6) Pg. 17 refers to the bases of the UJT backwards, though the diagram for the PUT is correct.
7) The square set of holes for Q6 is incorrect.

Mods I dreamed up (but haven't tried):

1) Rocker Bypass: Use a DPDT in the rocker pedal to switch from using the rocker pot to a manual pot set to the sweet spot for "yoy." Swap the two non-grounded legs of the pots. An LED can be inserted in series from the "cold" lug of the pot to ground to keep the wiper from grounding out.
2) Speed mod: Increase C2 and/or the Rate pot to increase the range of available speeds.

[R.G.]

Saved in the Ludwig design folder.

[Keppy]

While preparing for some oscillator speed muckery, I noticed something about Q1 today. With the animation rate set to full speed, and R4 set to zero resistance, animation toggle set to fast, and animation turned off, the transistor has a 35v drop from collector to emitter with no external current limiting on the collector or emitter. I was thinking it was possible to fry Q1 with basic (albeit unlikely) user error, since the base is tied to +35v. Then I realized that the base current is extremely low (~30uA) and hfe tops out at 250 for a 2n5551, limiting collector current to <8mA. Um, right? Also, 8mA*35v makes 280mW dissipation, well below the 625mW power rating.

So Q1 is safe at all settings... right?

[R.G.]

I'll go look. May take a couple of days. I'm kind of swamped with regular life right now.

[pinkjimiphoton]

hi guys,
yeah, i did come up with using an ev5 expression pedal with a little modification... i can swell PAST the filters into what sounds like straight fuzz.. very cool.

the quicklock universal volume pedal works right out of the box almost perfectly. personally, i don't like the crybaby with this, it doesn't "feel" right... but the extended sweep ya get with an EXP is kinda cool... heel back you can make it self oscillate at whatever pitch you end off playing, and toe down you can effectively bypass the filtering. very cool.

i'll try and shoot a little video of it, and the proper brand exp pedal.

i, too, prefer the clone to the original. i like the clone so much i didn't mind passing on cousin lewdvig to a new guy... and it sold for enough to pay off all my parts and pedal addictions.... and get a couple (ok, like, 14) more guitars.

i think where the pedal was concerned, it was supPOSEd to act like that... it makes sense, you can swell from fuzz guitar to synth sweeps. i love this thing!

big props to brother Dino, for building mine for me. thanks again bro!! i just can't see enough detail to do anything that complex anymore, tho having the proper bifocals now helps a lot.
be interesting to see this whole mess with all the mods and improvements implemented! 

[Govmnt_Lacky]

i just can't see enough detail to do anything that complex anymore, tho having the proper bifocals now helps a lot.
be interesting to see this whole mess with all the mods and improvements implemented! 

So... I guess this means you are ready for the SMD Ludwig then Jimmy?