usekgb
Well-known member
JohnRoberts said:
While a microprocessor is cheaper and easier these days, there is something desirable about the sound of old BBD technology. It has many flaws, but it sounds really nice when implemented well.
MXR Flanger clone
- Thread starter Spencerleehorton
- Start date
Help Support GroupDIY Audio Forum:
While a microprocessor is cheaper and easier these days, there is something desirable about the sound of old BBD technology. It has many flaws, but it sounds really nice when implemented well.
Back at the time the extra distortion was not desirable enough for people to stay with BBD delay over the more clinical early digital delays that were also far from perfect themselves. I know, I was in those trenches.usekgb said:
As I mentioned there were technical papers written specifically investigating the characteristic distortions in charge coupled devices. Not a simple non-linearity, since the distortion per stage gets imprinted 500 or 1,000 times for a single pass through the shift register. Not to mention some delay/phase differences with global distortion mechanisms printed to all stages in real time but receiving different delays when finally output.
In my judgement the fact that BBDs are still in production is mainly to support guitar pedals. The customer is always right even when....
JR
Spencerleehorton
Well-known member
Ok seems like we are getting somewhere now, if the mn3207 is a good replacement I will order one and start printing Pcb.
I was thinking about having a daughter board which will connect to the 16 pin connector which will be on the main Pcb so I can try the sad1024 to compare.
I was thinking about having a daughter board which will connect to the 16 pin connector which will be on the main Pcb so I can try the sad1024 to compare.
usekgb
Well-known member
JohnRoberts said:
I totally agree. While I would never use a BBD to run delay towers in a PA rig, they sound wonderful in guitar effects and reggae. The inherent non-linear distortion and limited bandwidth just sounds cool. I even use my analog tap delay on an effects send when I do sound for some acts.
As for a flanger, the same arguments can be made. Digital technically is "better", but analog is what we are used to hearing, and our ears are trained to hear it. Both technologies get the job done, and it becomes a matter of taste.
BBD delay was really limited for frequency response and signal to noise beyond more than tens of mSec. I added companding noise reduction to improve the S/N but there was no remedy for sampling theory related lack of HF content. In fact we needed to roll off some LF too to keep it natural sounding.usekgb said:
Digital delay pretty much drove BBD delay from the market.
Warning it's about to get all esoteric in here. 8) Flanging is a special effect where distortion can arguably be desirable in some situations (extreme range of flanging). If you look at the origin of flanging, originally known as "reel" flanging, because the short delays were generated from running two tape machines at roughly the same speed, and the leading machine was slowed down to match by dragging a hand on the outside of the tape reel.
Flanging basically subtracts (or adds) a delayed version of the signal from (to) the dry original, creating a series of bumps and notches from comb filtering. As the delay approaches 0 mSec, a clean identical delayed signal would cancel completely leaving nothing. :'( Flanging between two tape machines had enough random distortion and response errors that they would never cancel perfectly leaving a rich extreme flanging sound.
The less than clean or flat BBD flanging, like tape paths, did not cancel perfectly either, and while S/N could get better in BBDs at high clock frequency (less time for signal to leak from the buckets) even the best BBD flanger never goes to 0 mSec delay so doesn't completely cancel out. In addition to that any distortion in the delayed path will not cancel and be emphasized at the extreme shorted delays.
I will grant that for flanging BBD path distortion "might" be desireable. I suspect even it could be effectively mimicked in DSP to a fair approximation. Maybe even tape reel flanging , but people probably don't know what that sounds like (Itchycoo park?)
https://www.youtube.com/watch?v=3feplJn5ZHk
or even earlier from 1959 https://www.youtube.com/watch?v=2j0wfKMgSME
BBD was a much later 1970's technology
JR
PS: My preference for shorter BBD (1/2 1024 or 512) was for this hairier (technical term) sounding flanger at shorter delays. Note: you can also make it more extreme by recirculating output back into the input , not very easy with old school (tape reel) flanging.
Spencerleehorton
Well-known member
ok so lets get back on track, what volatges should i use if im using the MN3207 chip?
it says typical at +5v on the SAD1024 its 15v. I dont know what current its running at other wise i could work out voltage drop? 100mA? 200mA?
Also where is the bias on the MN3207 chip? and what voltage would it need?
ideally need to talk about how to hook this MN3207 into a SAD1024 slot, what are the pin configs?
regards
Spence.
it says typical at +5v on the SAD1024 its 15v. I dont know what current its running at other wise i could work out voltage drop? 100mA? 200mA?
Also where is the bias on the MN3207 chip? and what voltage would it need?
ideally need to talk about how to hook this MN3207 into a SAD1024 slot, what are the pin configs?
regards
Spence.
usekgb
Well-known member
If you want to wrap your head around the MN3207, read through this page: http://www.electrosmash.com/mn3007-bucket-brigade-devices
There is tons of info on this chip, and bucket brigade technology in general. Honestly, I would abandon trying to make your circuit compatible with the Reticon chip, as they are long gone, and very expensive to obtain.
The bias is on the input of the chip. On the 3208, it is pin 3. Check the data sheet for the 3207 to see if it is the same. The bias on the BL3208a usually lands right around 5.5V, but you will probably want to trim this value, and observe the output on an oscilloscope to make sure you are bossing correctly. You also need two clock inputs, which are opposite in polarity. You can use a BBD clock driver chip, or something else that will get you this.
There is tons of info on this chip, and bucket brigade technology in general. Honestly, I would abandon trying to make your circuit compatible with the Reticon chip, as they are long gone, and very expensive to obtain.
The bias is on the input of the chip. On the 3208, it is pin 3. Check the data sheet for the 3207 to see if it is the same. The bias on the BL3208a usually lands right around 5.5V, but you will probably want to trim this value, and observe the output on an oscilloscope to make sure you are bossing correctly. You also need two clock inputs, which are opposite in polarity. You can use a BBD clock driver chip, or something else that will get you this.
Spencerleehorton
Well-known member
Now I'm confused, your saying one minute that the chip is a replacement for the sad1024, now your saying don't bother trying to implement it in the 16 pin?
I don't know anything about this sort of thing but already I'm getting very confusing information, from the schematics I've seen which use the 8pin chips (3007,3207) it's seem they add a transistor on the output and a cap is reduced in pf a little.
Is it just these changes you are referring to, or is there some other changes?
So far I've seen a lot of schematics and talk about this but no definitive results of these changes, as most of this has ended up in schematics being drawn up and when tested they don't work.
I don't know anything about this sort of thing but already I'm getting very confusing information, from the schematics I've seen which use the 8pin chips (3007,3207) it's seem they add a transistor on the output and a cap is reduced in pf a little.
Is it just these changes you are referring to, or is there some other changes?
So far I've seen a lot of schematics and talk about this but no definitive results of these changes, as most of this has ended up in schematics being drawn up and when tested they don't work.
usekgb
Well-known member
Sorry for the confusion. The MN chips are a replacement for the SAD chips, but they are not a drop in replacement. There really is no drop in replacement for the Reticon chips, so some reengineering is required to use the MN series chips. Biasing is a little different, as well as power supply requirement. I think the MN3007 is now available again from CoolAudio, but I have not used it. Did you read the link I gave you thoroughly? Read it a couple times for everything to sink in. I'm on my phone, so can't really give you an exact answer on what to do, but you should be able to make the conversion using data sheets and the electrosmash information.
This stuff can be kind of tricky if you have never used BBD devices before, but it's not as complicated as you might think it is at first glance. Keep plugging away, and you'll get it.
This stuff can be kind of tricky if you have never used BBD devices before, but it's not as complicated as you might think it is at first glance. Keep plugging away, and you'll get it.
They are not that same chips.... I've already said this the SAD1024 has two separate 512 stage ASRs inside. The MN3207 has one 1024 ASR inside. It will only be functionally equivalent for that specific application because the MXR flanger ran the two halves of the SAD1024 in series.Spencerleehorton said:
No the outputs of the BBD delay are already MOS source followers inside the chip. Some cheap designs may add a transistor configured as a LPF, there can be substantial clock frequency component, if you don't trim for the output buffer Vgs. You probably won't follow this but BBD ASR only pass one sample of signal per complete clock cycle (one half period up one half period down). This sample is only valid for 1/2 of the clock period. To present a smoother output waveform the BBD has a 1/2 stage delay built in to feed a second buffer, so the one output sample per clock gets presented by one mosfet follower for one half clock, then the other mosfet follower the other half clock period. Since these mosfets will not have identical Vgs, there will be a small clock waveform superimposed on the audio output. Premium designs add a trimpot between these two source outputs to trim out that DC clock component. Cheaper designs just ignore and hope to filter it out with the anti-imaging filters typically used to reconstruct the output from time-sampled wave streams. (did you follow that?)
Look at the data sheet for the part you plan to use (3207). It will provide the information you ask, like range of operating voltage it will work over, ballpark input bias.
The audio signal being passed internally will need to be biased at a DC voltage some fraction of the total supply voltage that delivers maximum clean signal. In professional designs we used a trim pot there and looked for symmetrical clipping on the output when optimally adjusted. BTW this optimal bias voltage can change with clock frequency so maybe check at both clock frequency extremes to pick the best compromise. Letting the HF clock end be a little dirty will make for richer flanging but the 1024 long line will not be as short as 512 used in micro flanger or studio boxes so less complete cancellation at short delay end.
The prototype BBDs I got to work with in the early '70s not only had a bias shift with clock frequency, but would cause a DC output level shift with clock frequency that could be problematic for fast changing clock applications. For slow changing clocks (like a flanger should be) this changing DC will be LF and below the passband LF cut off.
Cheap guitar pedals will not likely trim clock noise or trim input bias voltage, but you will at least need to bias up the different maker BBD chip to a different nominal input bias. The odds of the two chip processes from different manufacturers decades apart being the same are extremely unlikely (do you feel lucky?).
The clock commonly used for these was a CMOS D-type flip-flop configured as a divide by 2 so 50% duty cycle was guaranteed. I suspect the same clock as used for the SAD will work for MN part. Note the clock chip needs to be connected to the same +V as the BBD. A higher voltage clock than the BBD is connected to could unintentionally bias on some internal parasitic junctions and damage the BBD.
Read the data sheets, maybe dig up some modern clone schematic already using the MN parts for some other ideas to copy.
JR
Spencerleehorton
Well-known member
Ok, well some of that I understood and some of it I didn't.
So I've ordered a MN3207 so that will be what I will first attempt to get working.
I have to say I agree that the Ada flanger sounds the best but there seems to be already solutions for making this and I want to start with the mxr flanger from this:
http://www.shredaholic.com/mxrflanger.html
Could we talk about how to implement the mn3207 within this schematic please?
So I've ordered a MN3207 so that will be what I will first attempt to get working.
I have to say I agree that the Ada flanger sounds the best but there seems to be already solutions for making this and I want to start with the mxr flanger from this:
http://www.shredaholic.com/mxrflanger.html
Could we talk about how to implement the mn3207 within this schematic please?
Attachments
Spencerleehorton
Well-known member
Apparently it draws 36mA so from 15v down to 5v I would need a 270ohm resistor.
Just would like to list the other changes that I will need to make.
Just would like to list the other changes that I will need to make.
midwayfair
Well-known member
Spencerleehorton said:
Again, people have already done this work for you:
http://lectric-fx.com/wp-content/uploads/2015/07/FlintlockFlangerV1.2.pdf
You also need to look at the datasheets for the chips and compare the pin functions.
I also highly recommend using the MN3007 instead of the 3207. The CoolAudio chips have issues and require more work to get them to sound right, and with the MN3007 you won't have to do anything with your 15V power supply except regulate it (if necessary). Unless you're building a bunch of these, the 3207 has enough problems to make it worth spending a few extra bucks on a better chip.
Spencerleehorton
Well-known member
Unfortunately that pcb isn't available, but at least I know that it's all possible now.
I will start comparing the differences in the mxr to the Ada version to see which one is best to suit my needs.
I will start comparing the differences in the mxr to the Ada version to see which one is best to suit my needs.
midwayfair
Well-known member
Spencerleehorton said:
I was more linking for the schematic, but you could e-mail them and ask when if they have a restock planned, or put a WTB on the Madbean forum. Despite the PCB being sold out I haven't seen a ton of builds posted, so there are likely to be some extras.
Andy Peters
Well-known member
You guys really don't want to know what we go through setting up a CCD image sensor with 10.5k x 10.5k active pixels (yes, 110 Mpixels) and sixteen outputs ...
Andy Peters said:
I had a 256x2 CCD array on the bench back in the early 70s but it was completely by accident (ICs were provided in open top packages, and the bench light trashed the audio signal.) ;D ;D ;D
JR
Spencerleehorton
Well-known member
Ok, I'll order a mn3007 and look into getting it going. Using the 4013 with it ok?
You realize the MN3007 is a P-channel device , not N-channel like the SAD1024? I do not have experience with either 3007 or 3207 parts since I moved on from BBD long before these variants came out (I used a MN3001 in my old kit). You can look at the data sheets for each part to compare S/N and distortion, if the data sheets are honest (?). Noise should be specified at the same clock frequency to be comparable, higher clock frequency will measure quieter.Spencerleehorton said:
I've actually used a P-channel BBDs with + supply but you need to take pains to make sure the + supply is very quiet so it does't corrupt the audio. The output stage resistors from the source followers that would be pulling up to ground when using a proper negative supply are now pulling up to +V, so any AC ripple on that +V supply can end up in your audio path as hum.
JR
Spencerleehorton
Well-known member
Still trying to get to the bottom of why the output is so low on this circuit?
R7 I'm thinking of upping to 150k to give unity but circuit seems to not be any different when I turn trimmer up to 150k?
I've put a 200k trimmer in R7.
Could anyone suggest anything to find why this circuit is a bit dull and low output when effect engaged please?
R7 I'm thinking of upping to 150k to give unity but circuit seems to not be any different when I turn trimmer up to 150k?
I've put a 200k trimmer in R7.
Could anyone suggest anything to find why this circuit is a bit dull and low output when effect engaged please?
