Hi all,
After a request for a schematic in another thread (https://www.diystompboxes.com/smfforum/index.php?topic=121377 (https://www.diystompboxes.com/smfforum/index.php?topic=121377)) I've been reading the schematics and doing a bit of analysis of the thing. I'll post the results on my website when I'm happy with it, like I did with my Boss MT-2 Metalzone analysis (https://electricdruid.net/boss-mt-2-metal-zone-pedal-analysis/). Seems like I find these weird 1980's metal pedals most interesting to look at - they're certainly not your standard overdrive.
The PDS-1550 is a odd thing - here's the structure of it:
(https://electricdruid.net/wp-content/uploads/2018/11/PDS-1550-Diagram.svg)
There's one bit I could use some help with though - the Channel A "metal distortion" stage. It looks like this:
(https://electricdruid.net/wp-content/uploads/2018/11/MetalDistortion.png)
Now, the thing that grabs my attention as most interesting straight away is the way that gain pot is wired so that there's a change in the first transistor stage Q3/2N5088 at the same time as a reciprocal change in the op-amp gain stage U11A.
I was expecting to see some tonal changes in the first stage Q3/5088 when the pot is altered, but in the simulation, I don't see anything. The pot seems only to change gain, which is a bit dull and could have been done more simply. So what am I missing?
There's some interesting features in this circuit though - asymmetric clipping around the op-amp, followed by a pair germanium diodes that I think act as a noise gate by not conducting until a certain threshold voltage is reached, followed by clipping diodes to ground.
So my question is: what's going on with those first two transistor stages, and why the odd arrangement of gain pot?
Doesn't it work as a LPF for the signal comic from Q3?
Sorry, I'm not understanding you. Does what work as a LPF for the signal from Q3? The string down to ground R6/C10/P6? Or the stage around Q7/2N3906?
I was expecting some tonal changes on the signal from Q3 from the changing resistance of P6, but I haven't managed to model it with the simulation yet. I get a lowpass response, but it doesn't change when you tweak P6. Probably my error.
For what i see, the role of Q3 is just to control gain.
The upper transistor is a booster, and the opamp is working as a non inverting stage.
Q3/p6 are there to simultaneously lower the opamp gain and the input gain into Q7, without loading the input, so you get a clean boost from q7 when turning down the gain. If the circuit was just Q7 and the opamp, if you lower the gain on the opamp, Q7 will be still clipping ( in an ugly way, i suppose)
That R18/R19 biasing network is a little odd. Should I assume it is being used to produce a bias voltage well ahead of the BBD used to produce the delay?
And DOD/Digitech regularly used gyrators in their designs to produce resonant bumps ahead of clipping stages. That's not one of them, is it?
Indeed, that bias is very confusing.
This stage is followed by the delay, but that R18/R19 bias network is separated from it by two different 1u caps (C13 and C14) so I don't think that is it.
I wondered if it might be to correct a DC offset caused by the assymmetric clipping.
It's also connected directly to the PNP's collector. What effect does it have there?
R18 is the one that confuses me. Otherwise the Q7 PNP stage is exactly the same as the Q3 NPN stage except upside down. R19 is 10K like R9, 470K and 120p to the base in both cases, 100K from there down (or up). Only difference is R13 is 10R and R16 is 100R, but that doesn't seem like a huge deal.
What's the purpose of having one NPN stage and then one PNP, do you think? Why would you do that?
PS: This is the full schematic if you feel the need to have a look at the whole thing and see how this section fits in.
https://electricdruid.net/wp-content/uploads/2018/11/Digitech-DOD-PDS1550-Programmable-distortion.pdf (https://electricdruid.net/wp-content/uploads/2018/11/Digitech-DOD-PDS1550-Programmable-distortion.pdf)
Quote from: ElectricDruid on November 20, 2018, 09:53:12 AM
What's the purpose of having one NPN stage and then one PNP, do you think? Why would you do that?
Signal polarity. Invert, then invert again.
(https://i.postimg.cc/sXFWDhJZ/korg1.gif) (https://postimages.org/)
R19 is the collector load for the PNP, as is R18, but the combination makes the bias to the op amp vary from 4.5v. I suspect this has something to do with the different clipping points of the diode combinations on the op amp.
regards, Jack
Boss HM-2 ;D
Yes, it's a somewhat complex clipping scheme. First two stages do not actually contribute anything significant - below about 40mVpp input signals. This seems like somewhat low figure but to put it to context, the input sensitivity of this circuit is so high that the shunt clipping diodes in the ouput will clip already with 4mVpp input signals and those asymmetric "soft" clipping diodes in feedback loop a bit earlier, with 2mVpp input.
So basically the second gain stage starts to asymmetrically compress the signal somewhat before the shunt clipping diodes begin to "harder" clip it symmetrically. Yes, germanium diodes are a "coring" noise gate.
So, contribution of first two stages to distortion is negligible. They do introduce slight amount of bandwidth reduction before distortion but compared to "modern" distortion designs the reduction is actually quite modest. Also the gain control scheme is designed to operate at pretty much full effective bandwidth of the guitar. Control's interaction with frequency response is therefore minimal and likely inaudible. That's the way the effect was designed. You can change it by altering the coupling capacitances, as usual.
I believe the purpose of the "dual gain control" is to achieve somewhat logarithmic tapering function while using linear taper because gain (or attenuation) in total becomes multiplied.
The bias network is not the ordinary resistive divider to "haf supply". If you look at it, R19 is basically a collector load for a feedback bias amplifier that automatically biases its collector output pretty close to half supply voltage. I think R18 is a "fine-tuning" addition to make sure that collector voltage actually remains around 4.5 volts. Somekind of "pull-up" resistor perhaps?
Quote from: amz-fx on November 20, 2018, 10:45:15 AM
R19 is the collector load for the PNP, as is R18, but the combination makes the bias to the op amp vary from 4.5v.
A quick look at the circuit in MultiSim reveals that the 62k (R18) doesn't do much. It reduces gain by around 1 db and changes the bias by little more than 0.1v. I suspect that the circuit would sound very much the same with or without R18.
regards, Jack
Teemuk is dead right about the Boss HM-2. They're virtually identical. Compare and contrast!
(https://electricdruid.net/wp-content/uploads/2018/11/Boss-HM2-Heavy-Metal.png)
Any idea which came first? They're both a similar era. To be fair, they both lift the first gain stage from the Big Muff Pi, and then do a flipped PNP version (that's still the odd bit) and then follow up with a Tubescreamer clipping stage, followed by DS-1/RAT diodes to ground. So there's nothing that's new in the world, just new c
ocktails to mix with existing ingredients (Hic!)
Quote from: amz-fx on November 20, 2018, 10:45:15 AM
Quote from: ElectricDruid on November 20, 2018, 09:53:12 AM
What's the purpose of having one NPN stage and then one PNP, do you think? Why would you do that?
Signal polarity. Invert, then invert again.
Sorry Jack, I don't understand this. The first Q3 stage is inverting, so why can't you just repeat another stage with an NPN to re-invert the signal. Why does it need a PNP instead?
Thanks,
Tom
Quote from: ElectricDruid on November 21, 2018, 10:32:43 AM
The first Q3 stage is inverting, so why can't you just repeat another stage with an NPN to re-invert the signal. Why does it need a PNP instead?
That is certainly the question. Two NPN stages should have worked just as well to restore the polarity to non-inverted. I suspect that Korg was just copying Boss, but I haven't a clue why Boss did it that way. They did have a tendency in older designs to want to use NPN followed by PNP stages, which can be seen in some of the digital delays where a 2-pole NPN transistor filter is followed by a similar PNP transistor lowpass (RDD-10 for example).
(https://i.postimg.cc/6pddfGGf/RDD10.gif) (https://postimages.org/)
regards, Jack
Supply noise cancelation maybe?
> A quick look at the circuit in MultiSim reveals that the 62k (R18) doesn't do much.
Did you put a real opamp after it? Without the 68K the opamp pin is slammed to its V-. Some opamps make nasty noises when you do that.
> a tendency in older designs to want to use NPN followed by PNP stages, .... NPN transistor filter is followed by a similar PNP transistor lowpass....
That's actually sensible. Direct-coupled emitter followers, all same parity, you lose 0.6V each stage. 4.5V bias at the first comes out 2.7V bias at the third stage. Flip polarity in each stage, you can go on forever.
Ah yes. Thanks Paul.
Still, it's a bit odd here, since they only use two stages (so the drop in bias level wouldn't have been a big deal) and then they mess with the bias with that 68K resistor anyway.
It'd have maybe made more sense in the Delay filter part of the circuit, where they've got several filter stages in a row.
I've put the analysis online now too. Any comments or corrections are appreciated.
https://electricdruid.net/digitech-pds-1550-distortion-pedal-analysis/ (https://electricdruid.net/digitech-pds-1550-distortion-pedal-analysis/)
I think it's become a bit long, so I'll probably keep working on it to try and make it a bit more concise. It's easy to write too much, and much harder to say the same thing in less words. But it makes for much better reading if you can!
Quote from: PRR on November 21, 2018, 07:02:15 PM
Did you put a real opamp after it? Without the 68K the opamp pin is slammed to its V-. Some opamps make nasty noises when you do that.
I used the LM833, only because it was the first bipolar opamp model I saw when I was adding it to the sim.
Quote from: PRR on November 21, 2018, 07:02:15 PM
That's actually sensible. Direct-coupled emitter followers, all same parity, you lose 0.6V each stage. 4.5V bias at the first comes out 2.7V bias at the third stage. Flip polarity in each stage, you can go on forever.
For the direct-coupled emitter follower that's true... probably not the best example since the gain stages in the HM2/PDS1550 distortions are AC-coupled.
Compare the DOD FX56 distortion circuit from about the same time, probably also copied from the HM-2.
(https://i.postimg.cc/XGvndsG0/DOD-FX56-American-Metal.jpg) (https://postimg.cc/XGvndsG0)
regards, Jack
That DOD FX56 schematic is exactly the same as the PDS-1550, which probably is to be expected. They bundled the Metal distortion channel up into a pedal in its own right, or perhaps they used the existing FX56 design for the Metal Distortion channel when they wanted a design for the PDS-1550. One or the other.
What I find interesting is that no-one seems to be aware that these pedals are direct HM-2 copies. The Boos HM-2 is getting more and more expensive and in certain circles is one of those "legendary" pedals. And yet there are lots of cheaper vintage pedals with basically the same circuit, even down to the hopeless errors which I wouldn't have expected from Boss.
A nice little detail of that design is that tone control's filters mimic the archetypal frequency response of a tube amp driving a reactive loudspeaker load: One peak in response at typical resonant frequency range, second peak to imitate treble boost due to rising coil impedance at higher frequencies. Metal Zone has a similar feature as well.
I'm reading the article, i love your analysis !
Just point out that the opamp stage is non inverting, though.
Quote from: dschwartz on November 22, 2018, 01:07:11 PM
I'm reading the article, i love your analysis !
Just point out that the opamp stage is non inverting, though.
Well spotted! Thanks.
If you see anything else...
T.
Quote from: ElectricDruid on November 22, 2018, 01:46:13 PM
Quote from: dschwartz on November 22, 2018, 01:07:11 PM
I'm reading the article, i love your analysis !
Just point out that the opamp stage is non inverting, though.
Well spotted! Thanks.
If you see anything else...
T.
Well..on the OD section, you mention that the clipping is "softened" by the series 10k before the shunt clipping dioded.. i think 10k will make the diodes clip pretty hard. Lower resistance makes the shunt clipping softer.
On the first page of the factory schematic, the H1A and H1E designations are reversed. H1A is the signal input and H1E is the power input.
Also, on the last page, it seems to indicate that the H1A (signal) connects to the electronic bypass switch control. I suspect that is incorrect and the bypass designation should be H1D on the last page.
Finally, isn't the input signal always mixed into the output via R74? The DC through this resistor from U1A also provides the bias for the output buffer.
If H1D is the bypass, since it blocks the parametric output from going into the buffer, then what turns the Overdrive channel On/Off? Q10 and Q11 are active at the same time and serve to isolate the Metal channel from going to the parametric. How is the Overdrive channel disabled? Or is it always going to the output (in parallel with the Metal channel) when the effect is enabled?
U3A and U3D seem to be serving the dual purpose of enabling the parametric and driving the channel indicators and switches.
I think the DIP switch SW3C disables the parametric at all times and not just for the Metal channel. (The para never seems to be engaged for the Metal channel, even with SW3C open, since the channel selector is driving it) Perhaps a more detailed look at the switching would be beneficial.
Given that there seem to be some errors in the factory schematic, I could be way off here! :)
Best regards, Jack
Quote from: dschwartz on November 23, 2018, 07:26:51 AM
Well..on the OD section, you mention that the clipping is "softened" by the series 10k before the shunt clipping dioded.. i think 10k will make the diodes clip pretty hard. Lower resistance makes the shunt clipping softer.
Thank you, you're quite right. I was remembering it the other way around, but it's not like that. I'll correct it.
This is *exactly* why it's important to get some other sets of eyes to have a look at something...
It's much appreciated and makes for better work. Thank you.
T.
Quote from: amz-fx on November 23, 2018, 09:35:49 AM
On the first page of the factory schematic, the H1A and H1E designations are reversed. H1A is the signal input and H1E is the power input.
Also, on the last page, it seems to indicate that the H1A (signal) connects to the electronic bypass switch control. I suspect that is incorrect and the bypass designation should be H1D on the last page.
Yes, I'd also spotted this. The "connections" between the "Jacks and switches" schematic sheet and the two pages of the main schematic don't seem to line up much at all!
Quote
Finally, isn't the input signal always mixed into the output via R74? The DC through this resistor from U1A also provides the bias for the output buffer.
Yes, I thought so. There's always a path from the input to the output. The FET switches only take the other signals out. It must be a nightmare to have anything like reasonable levels between the "bypassed" and "effect on" signals.
I've not dealt with any of the switching in my article. It just seemed like a whole other can of worms to open, so I just avoided it. I realise that means there's quite a bit left out. QUITE a bit. Another day, perhaps.
But I should mention just how much there is that I've not mentioned, if you see what I mean. The article comes across as pretty complete, and in fact, it's only really a first try. And the routing craziness is a big part of that.
Quote
If H1D is the bypass, since it blocks the parametric output from going into the buffer, then what turns the Overdrive channel On/Off? Q10 and Q11 are active at the same time and serve to isolate the Metal channel from going to the parametric. How is the Overdrive channel disabled? Or is it always going to the output (in parallel with the Metal channel) when the effect is enabled?
I think it is always enabled. This is another bizarre design decision in this pedal. By the time I got to this point, I wasn't even surprised any more. It's weird, but that seems to be what it does. The only FET switch between output of the overdrive stage and the output buffer is after the parametric EQ, which I assumed is to turn the whole effect off. To me that means there is no way to turn the overdrive stage off.
It seems we've always got clean signal, and we can then add the effect signal, and that effect signal is either just the Overdrive channel, or both the Overdrive channel and the Metal channel mixed together.
It's a truly odd pedal when you get into it. That's why I find this stuff so fascinating. I'm really struggling to reconstruct the thought process that made people think that was a good idea!! I suppose you'd have to mod it to separate out the parts and then you could see. Perhaps it sound hopeless with each bit on it's own!
Quote
U3A and U3D seem to be serving the dual purpose of enabling the parametric and driving the channel indicators and switches.
Agreed, they do.
Quote
I think the DIP switch SW3C disables the parametric at all times and not just for the Metal channel. (The para never seems to be engaged for the Metal channel, even with SW3C open, since the channel selector is driving it) Perhaps a more detailed look at the switching would be beneficial.
Hoho, yes, "a more detailed look would be beneficial". For me, that's a technical way of saying this part confused the hell out of me! ;)
I thought perhaps the DIP switch only disabled the parametric when the channel selector driving it was set to the Metal channel, but I haven't worked it through thoroughly.
Quote
Given that there seem to be some errors in the factory schematic, I could be way off here! :)
I think you're more right than they are, Jack, TBH...
Thanks,
Tom