Design vs Manufacturing (analog)

[J. Dan]

There are many on here who have more experience both as users and manufacturers than me with respect to some of the widely honored vintage analog gear.

 

I would like to see if some of my assumptions can either be confirmed and maybe expanded upon, or denied with corrections and explanations. For the sake of background, my education was in engineering (Electrical, Engineering Management, Economics) and for quite a time I, as a hobby in my younger years, designed, prototyped, and occasionally built circuits of my own design. That fizzled out throughout the early 2000's as life took over. I did successful repairs on my own and friends' gear not for pay but just because I enjoyed it and had the time back then (90s). I got very interested in analog electronics and while all of the greatest vintage gear at that time had gotten in such demand and so expensive that I had little to no access to it, I did study schematics along with writing my own and do a lot of experimentation on breadboard. Only vintage analog stuff I ever repaired and/or calibrated was PolySix, Jupiter 6, and Opus 3. Not much, I know.

 

Ok, now for the question. To again clarify, I'm talking about the difference between a design on paper that may involve discrete components as an example, vs actual production where there has to be repeatability and concerns around things like temperature effects, effects of drift in line voltage, etc. Of course some vintage devices were notorious for tuning drift as they warmed up, and future products solved some of those problems.

 

So from a pure design standpoint, I can design an Oscillator or Filter with transistors, resistors, and capacitors. On paper, that would be the Moog Ladder filter. I BELIEVE (correct me if I'm wrong) they ended up using an integrated circuit that was just the transistors. Hold that thought. Later on, we have OTA's (Operational Transconductance Amplifiers). For those who don't know what that is, it's similar to the Op Amps used everywhere now -every single stage of a miser for example, except that you can use a current or voltage to remotely control the gain. Easy way to make something voltage or current controlled. Again, correct me if I'm wrong, but I believe the Korg MS20 filter was based around the CA3080 OTA. Another popular OTA is the LM13600. I have used both of these extensively to breadboard a lot of my ideas. Now fast forward to the Curtis and SSM chips. Dedicated Oscillators and Filters and VCA's....much more purpose build and expensive and now harder to get, while the aforementioned generic OTA's are still available. So why use them?

 

My ASSUMPTION is that due to the tighter controls and repeatability of the manufacturing practice, that getting more of it baked into the chip rather than relying on the tolerances of multiple external components from various supply chains resulted in a more repeatable and reliable end product off the production line. Is my assumption correct or not?

[mate stubb]

Ok, now for the question. To again clarify, I'm talking about the difference between a design on paper that may involve discrete components as an example, vs actual production

My ASSUMPTION is that due to the tighter controls and repeatability of the manufacturing practice, that getting more of it baked into the chip rather than relying on the tolerances of multiple external components from various supply chains resulted in a more repeatable and reliable end product off the production line. Is my assumption correct or not?

 

 

Yep. More discrete components means greater tolerance stack up problems. But it also means more board space, more assembly time, and larger parts inventory.

 

Economies of scale help to determine how integrated to get. If you are making a few of these items, it will generally not pay to invest in custom chips. Speaking of OTAs, I believe Dave Smith is using a somewhat generic quad OTA chip these days for the P6 filters. The original was a custom SSM 2040 chip which contained 4 OTA cells plus the associated caps and resistors.

[GRollins]

The short answer is yes, that you can get more repeatable results, etc. etc. etc.

 

But...

 

There's repeatable and then there's repeatable. The devil's in the details. You can execute pretty good resistors and very good transistors on a chip, but the fly in the ointment is the capacitors. They're just not up to snuff. From a high end stereo point of view (a Venn diagram would show some, but not complete overlap with keyboard/synth design), you'd like polypropylene, or (better) polystyrene, or (better still) Teflon caps in nearly every part of a circuit. Electrolytics and ceramics have their uses in certain applications, but for the signal path, you want good film caps.

 

For synth design...well...

 

Then you start getting into arguments like the original Minimoog used carbon comp resistors (terrible) and ceramic caps (terrible in the signal path), yet it "sounded good," meaning that we grew up hearing "that sound" and it's baked into our DNA. It's vintage. It's analog. It's history. It's fill-in-the-blank, but the long and short of it is that in the case of the Minimoog (and other designs) it became part of the Mojo.

 

And if you execute the same circuit with good quality Vishay metal film resistors and Teflon/tin foil caps...?

 

Honestly, I can see both sides of that argument, but I'd come down on the side of better resistors and caps. If nothing else, carbon comp resistors drift horribly with time due to humidity and you'll see unpredictable results ten or twenty years down the road. There may be subtle differences in the sound between the original parts BOM and a more modern one, but I can live with it. My position is that the sound differences between an original Moog and a New And Improved one would be swamped by unit-to-unit variations of the original as they came off the production line due to 20% or greater tolerances in caps and 5%/10%/20% tolerances in resistors.

 

Monte Carlo analysis, in other words.

 

But that still doesn't change the fact that caps in chips ain't up to stuff. Better tolerances? Yes. More stable over time? Yes. Better sound quaility? Er...no.

 

Styrene caps just don't fit into chips.

 

So I'd say go with something like, say, THAT Corp. matched transistor sets, but use decent external resistors and capacitors.

 

Grey

[The Real MC]

ICs do not always equate to better tolerance.

 

If you use the incorrect dielectric cap for the timing core of a CEM3340, it will still drift.

 

The famous Oberheim SEM/OBX filter relied on matched 3080 OTAs when used in a polysynth. The reason was better matching of resonant color between voices. While the 13600/13700 does contain two 3080 cores with added predistortion diodes and output buffers, the OTA pairs are NOT matched. The only "matched" dual OTA on a chip was the CA3280, which is long out of production.

 

The SSM2050 EG was a royal PITA for rev1/rev2 Prophet-5 manufacture because the EG transient times varied from chip to chip. SCI had to hand test and match sets of ten for each unit (5 for VCF EG, 5 for VCA EG). That's a major reason why SCI switched to the CEM chipset for the rev3, the CEM3310 EG had far matching of transient times between chips.

 

Sure the Minimoog used godawful resistors and caps in the filter. Sometimes imperfections can be a GOOD thing. I certainly am NOT re-capping my vintage RA Moog Minimoog just because newer caps are "better".

 

I am not a subscriber of the knee-jerk "every vintage gear needs a re-cap!" mindset. My OBX had dirty power rails. Took pictures of the scope traces before and after recap. There was NO difference, power rails were STILL dirty.

 

ICs do not always equate to better performance. ARP's discrete expo converter in their VCOs still rules for tuning stability with minimal components.

[GRollins]

Unless the caps were bad I wouldn't expect that a recap would clean up dirty rails. I'd look at the regulation. If there's no regulation, I'd check the diodes in the bridge. If they're okay, I'd start thinking that the designer used either too small a power transformer or too little capacitance in the power supply (likely both)--either would lead to too much ripple. Where to go from there is up to the owner. Protestations that [fill in the name of a famous designer] is a genius carry no weight with me. Famous people are just as likely to scrimp on power supplies as no-name people, either thinking that power supplies don't matter (a surprisingly common attitude) or that it's a good place to save money.

 

Whether or not to recap a piece of gear is something I approach on a case-by-case basis. I've got an old Hafler pro amp that's humming--clear candidate for caps. It's sitting on the floor here next to me, waiting for me to go through with calipers to work out the lead spacings so that I don't have trouble fitting new parts on the PCB. On the other hand, I've got old stereo amps about the same age that are silent as the grave. I haven't messed with them.

 

Carbon comp resistors bother me. It's bad enough to see a 20% tolerance part in any circuit, but to see a 5% or 10% resistor that's drifted 50% or more from its value over time? Uh-unh. That does not make me happy. It's not a matter of vintage vs. modern, it's a matter of "this thing doesn't sound the way it did the day it was made." Here in the southeast, the humidity creeps into the resistors and they are waaaaaay off.

 

I had a big box full of random value carbon comp resistors that a professor gave me back when I was in school. Leftovers from a circuit they had built for the lab. A mere five or six years later I started going through the box testing them, trying to match pairs for a filter I wanted to build. I could match 'em, all right. That wasn't the problem. The problem was that I could scarcely find any that were still within their rated tolerances. I ended up throwing the whole box away. Note that these were brand new (Allen Bradley? I don't remember--it was a good company.) 5% and 10% parts when I got them. By the time I wanted to build the filter, they were well on their way to being 20% parts. Ugh. That made a believer out of me. Carbon composition resistors cannot be trusted.

 

Grey

[J. Dan]

Would be also interesting to discuss as well as compare/contrast the modern crop of reissues.

[GRollins]

There are numerous comparisons of the Behringer Model D and the Moog. I imagine that the limited edition ARP2600 is being compared to the original somewhere--I haven't looked. Once the Behringer 2600 comes out, there will no doubt be three-way comparisons.

 

I know some people seem to be losing their noodles over the idea that the Behringer 2600 prototype doesn't seem to have a built-in spring reverb. Reverb as an effect is nice in its place, but it's not necessarily the first thing that springs (ahem) to mind when thinking about a 2600. That said, reverb--or the lack of it--will clearly change the sound of the instrument. Assuming that the Behringer unit is cheap enough, there's no reason not to add an external reverb, but that seems to be verboten, for reasons which I am unable to fathom. Not that reverb tanks are necessarily the same sort of thing as resistors, capacitors, transistors, and tubes, but they're a component.

 

Grey

[J. Dan]

Yes lots of side by side reviews. I'm speaking more about the use of modern components and manufacturing techniques and reduction not just in material cost but labor, capital expense, economies of scale, rework, etc.

[mate stubb]

Well, how about starting with surface mount vs thru hole components. I'm assuming the new stuff is surface mount and can nowadays be stuffed by a pick and place machine at a board house cheaper than doing it by hand.

 

But, I haven't actually seen inside any of the new analog stuff. Pretty sure Dave Smith uses SM.

[The Real MC]

Well, how about starting with surface mount vs thru hole components. I'm assuming the new stuff is surface mount and can nowadays be stuffed by a pick and place machine at a board house cheaper than doing it by hand.

 

But, I haven't actually seen inside any of the new analog stuff. Pretty sure Dave Smith uses SM.

 

Moog stuff (not the model D or modulars) is surface mount. Voyager and most of the MF pedals were THT.

 

You CAN make surface mount sound good. My Minitaur can easily compete with my original Taurus pedals.