Fake BBD's or Bad Design?

[vigilante397]

I'm working on a Deluxe Memory Man clone running at -15V like the original, which means it needs MN3005 like the original. Me being the cheapskate I am ordered some sketchy MN3005's from a Chinese seller on eBay. They arrived today, and needless to say they don't work. Here's what's going on:

With the chips out of the sockets (thank goodness I used sockets) My voltages are fantastic. Exactly what they should be (based on Madbean's project document for Total Recall). I'm getting -15V at VD, -7.5V or so at the clock pins and the input, and 0V on the output pins. When I  check with an audio probe I can hear the clock on the clock pins, which changes frequency as I move the delay pot, and I get beautiful clean signal on the input pin.

I started by plugging in the first MN3005 stage, as Madbean recommends, for biasing, and everything went to hell. The voltages looked like they started in the right ranges, but they all dropped rapidly and the chip started heating up FAST. When I poked with an audio probe the input pin had clean signal mixed with clock noise, and there was nothing but clock noise on the output pins.

So my question is, does this sound like the sort of thing that would happen with a bad chip (heating up that fast makes me think internal short)? Or does it sound like a design error somewhere along the line that could be frying a perfectly good chip?

[Rob Strand]

So my question is, does this sound like the sort of thing that would happen with a bad chip (heating up that fast makes me think internal short)? Or does it sound like a design error somewhere along the line that could be frying a perfectly good chip?

There was a thread about BBDs heating-up within the last two years.   IIRC the chips were crap.

I'll see if I can find the thread.

Here's one, see last post,
https://www.diystompboxes.com/smfforum/index.php?topic=116938.msg1084624#msg1084624

and another,
https://www.diystompboxes.com/smfforum/index.php?topic=119746.0

Something in the back of my mind says one of those posters posted a follow-up thread on the same issue a little after their thread finished.  I can't find it though.

[vigilante397]

Thanks Rob, didn't even think about searching, and embarrassingly enough I was part of that first thread there But the last post on the second thread is interesting, that some of those chips were just re-labeled MN3207's instead of MN3007. That would obviously explain the heating problem I'm having, as I would be supplying reverse-polarity voltage to the chip, and the chip probably wouldn't like that.

Aside from the fake MN3005's I also have some CoolAudio V3205's, and I was thinking of jumping header pins from my IC sockets to my breadboard and running those (after adjusting the voltage, of course). I may as well try that with these fakes as well. That will likely be a project for tomorrow, so I will report back on that

[Rob Strand]

MN3207's instead of MN3007

I'm not sure if they were re-labeled or they just didn't work at higher voltages.
(IIRC the power rails get flipped between the two.)

I shouldn't use my memory.  Yeah, he pretty much says they are MN3207's.

[Kevin Mitchell]

Larry (armdnrdy) once started a thread for a BBD tester. It worked for a few different ones and would even display the amount of stages detected. All without a microcontroller!

This thread
Sadly the photos aren't available anymore and I haven't seen Larry around in a while. I'd so build one just to have it! Especially considering how BBDs will likely become even more illusive over time.

But all of this wont help if you've got mislabeled chips! Those bastards.

[Mark Hammer]

Are the chips perhaps MN3205?  Same number of stages but different powering.

[vigilante397]

Are the chips perhaps MN3205?  Same number of stages but different powering.

That's the suspicion I mean I would be shocked if they were MN3207, I was just mentioning MN3007 vs MN3207 because that's what they were talking about in the threads Rob mentioned. I brought the board and the chips in question to work with me and I plan to jump the IC sockets out to a breadboard so I can play with the pinouts and see if the power connections are backwards. Also I got rid of the oscilloscope I had at home so it will be nice to use the one I have at work to check some things out if I can get the chips working.

[Mark Hammer]

Original MN3005 were 8-pin on a 14-pin die, with three empty pins on each side.  No MN3207 has ever been made that way, to my knowledge.  Even the recent thru-hole MN3005 reissues from Xvive have that form factor (although the SMT version is an 8-pin size).  The MN3205 and V3205 look like those 3005s.

[Kevin Mitchell]

And the 05 BBDs physically larger than the 07s so that's a dead giveaway. EDIT: THANKS MARK 
However there is the 08 model. I think some small versions of these were made labeled BL3208. But that's irrelevant and I'm rambling.
Here's a datasheet anyhow

I enjoy a good mystery.

[vigilante397]

No MN3207 has ever been made that way, to my knowledge.

And the 05 BBDs physically larger than the 07s so that's a dead giveaway.

I do not think these are MN3207 or MN3007, that was never my thought. I only mentioned those because the threads Rob linked to were discussing fake MN3007.

But moving on, I just did some wire poking. I'm using an SMPS to take a +9V input and go to a -15V output, but I swapped a resistor and it's now happily putting out -9V, a safe voltage for MN3205. I then ran jumpers from my IC sockets to a breadboard and swapped the VDD and GND pins, then put a series resistor between VDD and VG to drop the voltage a bit. Then I plugged in the "MN3005," and wouldn't you know it, it's acting a lot like an MN3205. Where before the chip was getting absurdly hot and the board was trying to suck my power supply dry, the chip is now running cool as a cucumber and the whole board is pulling a steady 38mA with just one BBD, and 65mA with both.

I don't have a guitar or amp with me, but with a signal generator and oscilloscope I can see a clean sine wave going all the way to the input of the first BBD, unfortunately I'm not seeing anything much on the output, and I can only get the sine wave to pass to the output by blending the delay out, so there still seems to be something wrong, or I just may have fried the chips yesterday, anyone's guess But I was able to verify the clock is working properly and the clock signal is getting to the BBDs, and I don't have any clock leakage to into my audio signal that I see, so at least I have some good news. But fried or not, these chips are most definitely not MN3005's, so I will be contacting the seller. I'll probably set these aside for now and since I have the breadboard set up for lower voltage and positive voltage I may as well try my V3205s to check my circuit before forking over a bunch of money for real 3005s. More work to do, and I'll keep everyone posted as it goes

[Kevin Mitchell]

I knew you weren't implying they were 07s just pointing out obvious stuff to further confuse folks 

It's a cruel world out there. My last ebay purchase was for a pack of BF245A and they didn't do a darn thing. It was grand. Weeks on end debugging session for a project with over 300 components 

[j_flanders]

unfortunately I'm not seeing anything much on the output, and I can only get the sine wave to pass to the output by blending the delay out, so there still seems to be something wrong, or I just may have fried the chips yesterday, anyone's guess

Are you supplying the correct bias voltage to the input signal?
A bit over 1/2Vdd is a good place to start.
Only a tiny range will produce an output, the rest will give no output at all.
Within that tiny range you need to find the optimal bias voltage that gives max headroom / symmetrical clipping.

[vigilante397]

Are you supplying the correct bias voltage to the input signal?
A bit over 1/2Vdd is a good place to start.
Only a tiny range will produce an output, the rest will give no output at all.
Within that tiny range you need to find the optimal bias voltage that gives max headroom / symmetrical clipping.

That's good to know, thanks! I tweaked the bias pot a little bit, but I did my typical thing when I don't know what I'm doing: start with it dead center, if it's not working turn it all the way CW, still not working then turn it all the way CCW. Doesn't give the best results if there's a specific window that it's supposed to be in. I've done a couple BBD choruses but this is my first experience with anything bigger than the 3007. So I'll play with the bias some more, maybe on my lunch break (unfortunately I can't spend all my time at work playing with pedals ) and let you know how it goes

[j_flanders]

You best probe right at the output of the first bbd, because they are in series and when the bias voltage of one bbd chip is off there will be no final output.
Maybe you can post Vdd, Vgg and Vbias of each bbd chip?

[vigilante397]

You best probe right at the output of the first bbd, because they are in series and when the bias voltage of one bbd chip is off there will be no final output.
Maybe you can post Vdd, Vgg and Vbias of each bbd chip?

I will do it

[j_flanders]

I don't have any clock leakage to into my audio signal that I see, so at least I have some good news.

With a proper layout you don't have clock noise leaking in the (dry) signal.

Clock noise is a problem in the delay signal because it's part of the bdd output signal.
You can reduce the fundamental clock frequency at the output of the (last) bdd.
There's a + and - clock frequency. One is the (phase)inverse of the other. You can make them cancel each other out by using identical resistors at the output of the bbd. More practical is using a trimpot instead of two separate resistors. The pot is usually labeled: 'balance trim'.

Apart from the fundamental clock frequency there's also other clock noise to consider.
It's similar to the sum and difference frequencies you get from (modulation) distortion.

Have an 8kHz clock signal? Feed the bbd chip a 3kHz sine wave or guitar note/harmonic and the output will be:
3kHz guitar note (delayed)
5kHz sine wave (from 8kHz clock - 3kHz guitar note = difference)
8kHz sine wave (clock frequency = fundamental)
11kHz sine wave (from 8kHz clock + 3kHz guitar note = sum)
There will also be the next lower and higher sum and difference harmonics: 2kHz, 11kHz, 14kHz, 17kHz and so on. But the higher order ones are lower in amplitude.
The first harmonic is the most problematic.

This is why the lowest possible clock frequency (longest delay time) is twice the highest input frequency.
Or vice versa, the highest input frequency is, at most, half the clock frequency.

That way you can filter out the first harmonic without throwing away the input signal.
For example:
With a 8kHz clock freq you can input 3,9kHz. The first harmonic will be: 8kHz - 3,9kHz = 4,1kHz.
In theory you could filter out that 4,1kHz and keep the 3,9kHz input signal. Though in reality filters don't have that steep roll off

But feed it 4,1kHz and the output would contain: 8kHz - 4,1kHz = 3,9kHz.
If you'd use a LP filter and filter out that musically 'unrelated' clock generated difference frequency of 3,9kHz, you'd also throw away your input signal of 4,1kHz.

I've done a couple BBD choruses

The delay times are very short, in the few or tens ms, and accordingly the clock frequency will be quite high (much higher than 20kHz or the audio range) and thus you don't experience the same problem there.

[vigilante397]

Brief status update: I hadn't updated in a while as I wasn't able to get my bias voltage for the first BBD anywhere near the ballpark it was supposed to be in, so I had tabled it for later. I finally got back to it last night, and I found the problem: I was missing a capacitor! Somehow as I was copying in the schematic I missed one C14 from Madbean's schematic was completely missing from mine, so I sliced the D_IN trace and added it inline, and lo and behold I'm getting good voltages! This was after midnight last night so I didn't get a chance to throw in a 3205 and see if I get output, but I expect that to happen today

[Rob Strand]

Brief status update:

So in the end were the chips  MN32xx, MN30xx or total duds?

[vigilante397]

So in the end were the chips  MN32xx, MN30xx or total duds?

Right, forgot to mention that. Looks like MN32005's. I contacted the seller and was given a refund.

[Rob Strand]

Right, forgot to mention that. Looks like MN32005's. I contacted the seller and was given a refund.

So basically same conclusion as the other threads.   
Oh well, at least they refunded you.