i seen in the 808 documentary that roland used damaged or defective transistors they got at a discount because they couldn't afford new ones and the roland founder claims that this gave the 808 its distinctive sound or "sizzle". then when the 808 became a hit they had no way of getting a hold of such transistors so they couldn't capitalise on its success.

what an interesting story, and another classic roland curse/twist of fate.

i wonder what would the machine have sounded like with the proper working transistors? is this why there are said to be discrepancies between some machines?

curious how can you use defective components and have stable results?

is this like a reverse volca beats snare situation?

Behringer will clone that defect

Interesting. There's a story that Peavey bought a huge batch of transistors that had been made for NASA. Once the batch ran out, the transistors could never be replaced like for like. They were straightforward 2N3055 type power transistors, but radiation hardened and with tightly specified gain, parasitic capacitances, etc.

Here is some information and photos of the transistor.

The rejected transistor at the heart of the iconic Roland TR-808.

I love his blog!

So how does the Yocto (and the eurorack 808 drum modules) get around this - what do they use?

Paging @ decoder23

Thanks, it's my blog! I haven't tried a Yocto, the couple of people who I have spoken to that have built 808 clones told me that they had to source an original RNZ transistor for their build. People look at the Roland schematic and understandably assume that it's a regular 2SC828-R and then find that it doesn't sound right. I to would be interested in how the Yocto approaches the snare and hand clap noise.

hopefully
or have an option to switch between defected / modern one.
but either way , if many 808's dont sound identical then i'm not too fussed.
it'll be compressed and tweaked at some point .

sorry but this sounds like a crock of ****.
do you have any evidence to back this up?
the difference in sound between the clones and the real 808 is less than the difference between two real 808s. have you even owned any of them ?

edit: oh, just saw the last paragraph. trying to sell Roland Boutiques, mystery solved.

Firstly I have no affiliation whatsoever with Roland. I tested the TR-08 and reported my findings. I have emails and messages from people who have tried to use other transistors in clone attempts and they have told me how poorly it worked out. I spoke with two of them today, they both had to source the original part. I have serviced many, many 808's in my service centre for decades and, most importantly, if you could be bothered watching the interview you will understand that it isn't me who is claiming that the TR-808 requires this specific part, it is the founder of Roland and designer of the TR-808. So believe what you like, I believe Ikutaro Kakehashi and my own decades long experience servicing these machines. I am very familiar with what a healthy TR-808 sounds like and what one with a defective RNZ sounds like.

I’m working on part two of the TR-808 transistor post, building a test jig with the 808 noise circuit to analyse the characteristics of the original noise selected transistor and to compare it’s level and spectrum to other transistors and even some zeners in order to try and help out DIY’ers and Yocto builders to choose a part that is closest to an original RNZ for their Q35 noise transistor rather than a standard 2SC828. Any suggestions for transistors and zener diodes to try out let me know and I will source them and do a comparison. Hopefully we can find a good solution for folks building their own TR-808 clones.

I got 808 defective transistors but a 2SC828-R ain't one

triez. 2N3904, BC548, 2SC945 (general purpose medium power NPN) or 9-13v Zeners. I'd imagine you may well find that different manufacturers devices produce different levels, spectrums and energy densities of noise, Multicomp devices may be noisier than Zetex for example (sorry Multicomp), due to slightly different fabrication processes. I'd also imagine you know all of this already.

You used to be able to buy specific 'noise' transistors. I used them in the late '70s to include in ADSR boxes for sound effects. If only I could remember the part numbers! ETI (Electronics Today International) ran a series of articles on DIY synths and went into noise sources in some detail. If you could find some of those old articles you might find some interesting info. I vaguely remember phrases like 'fine grass', 'long grass', 'hash', 'spitting' and 'fried breakfast'.

The Q11 transistor in the CR-8000 performs the same function and configuration as the part in the 808. This is a 2SC945P(NZ) - and a specific NZ (not sure what it stands for) part for that particular transistor which makes it different form the other 2SC945P in the same board. Given the circuit similarities between the CR-8000 and the TR-808 I would love to hear how close or different is the noise generated by these transistors.

There is a certain missing bright metallic quality to the TR-08 that the RD-808 seems to have done much better.

Are there any demos from the RD-808 with proper line recording? Cause otherwise it's impossible to tell.

The Behringer people themselves didn't want to make such recordings because the RD-808 is still a prototype, its sound and features aren't final.

i am using these

It will be very interesting when the RD-808 hits and people crack them open to see what Behringer's solution to the issue is. I suspect they will have leveraged their economies of scale and made their own part -- possibly a custom diode that replicates the noise behavior.

One thing worth noting that no one seems to have brought up -- Roland could have easily been making 808s for years after the fact but Kakehashi apparently said "no". This is significant when you take into account other Roland products that changed vital sound components midway through their production run -- Jupiter-4, VP-330, and MKS-80 as examples. Each continued production because the resulting sound was the same or close enough that it could still be called the same model. Sounds like this wasn't the case with the TR-808. However, I do think the explosion in the 808's popularity happened long enough after production finished that it wasn't economical for Roland to go back and make more.

Very true, and even if snare and clap wouldnt have sounded exactly the same, people werent that anal about small differences in sound back then.. i mean pre Internet without thousands of a/b comparisons and without having thousands of "experts" in forums? You can literally use any 828r and adjust 1 resistor to get VERY satisfying results.. that's just a nice anectote in that documentary, nothing else.

I do have the construction manual for the ETI synths, I actually gave one of those synths away earlier this year, from memory they used a digital noise generator, I will go and check. It will be an interesting journey trying out transistors and zeners, It would be good to find something closer to the original, fortunately I have 10 NOS RNZ transistors so I have a good basis for comparison.

Thanks for the heads up! I have a couple of scrap CR-8000's in the store room, I will remove the noise transistors and test them.

Roland did change the layout of the noise section in the 808 during production, you can see two different versions of the amplifier on the schematic. It's quite the detective story. If Behringer does ever bring out their version it will be fascinating to see whether they use a semiconductor noise source and if so how it's raw output compares.

triez. Yes, the Trancendent 2000 used a psuedo-random shift register circuit, but there were other DIY synth circuits and sound effects boxes that used a specific noise transistor. Of course, the great thing about a noise transistor is that it's a single TO92 package, a shift register circuit takes up a lot of board space.

Apart from Zener diodes you might try reverse biasing normal diodes, ones with high leakage and the lowest PIV. Keep them cold and dark to maximise noise.

Found it! The component I was thinking of is a MM5837. It's actually a psuedo random circuit in an 8 pin DIL package, so it is compact but certainly not as compact as I thought I remembered.

i'm not claiming Roland did not use a defective part, i'm saying it wouldn't make nearly as much difference in the sound as you claim it does.

I would ask for an audio demo so you can demonstrate this supposed difference, but i already know your ears are broken if you think the TR-08 sounds closer than a well-calibrated Yocto

Once I get the circuit built I will test plenty of transistors and zeners, I will post a spectral analysis of the original transistor too. It would be good to be able to help out the Yocto builders with some alternatives.

I suppose you could write to various manufacturers and ask them to provide samples of the hundred noisiest transistors they manufacture.

The 808 was laughed at. Other drum machines were playing samples of actual drums. It was never seen as a drum synthesizer but a relic.

I'm not sure that it's about the noise level, it's about the noise spectrum. It may well be that the RNZ actually has a low noise output compared to parts that other manufacturers were using for noise but it clearly had the sound that Mr. Kakehashi was looking for.

Building a replica with a different transistor will give a different sound so you won't have a perfect 808 snare or hand clap but that might be beside the point for DIY'ers if the sound of the component that you do use suits your taste. It's all part of the fun.

triez. Definitely, it's all about the timbre of the noise. I suppose that's why so many synths used shift register designs, they're more predictable and can be endlessly tweaked to adjust the spectral energy density.

I read one 'tech tip' in which a bloke started with 30 'identical' transistors and found 2 that suited his purpose. Flutter, spitting and 'fried breakfast' can all ruin a good prospect. I'd imagine your test circuit will need a fairly robust transistor socket!

I am really interested in this, actually started a thread about if anyone had a certain flavour of noise on synths/drum machines that sounded better or different from others.

I am no DIYer but i've heard enough to have heard some noise types that stand out as ones I like. i dont know if thats more the circuit around it or the noise but its defo there

good starting point would be 945p, they used "noise selected" ones in the tr606.. wondering why they didn't use the "fake" 828 in the 606? Imho, the 606 snare is at least as good if not better sounding than the 808 snare, but that's just personal taste of course

I agree, it's just as good and its transistor is a perfect candidate for clones. Also can be found in the CR-8000 as I said earlier, as this selected 945p seems to be exactly the same part as in the 606. Which begs the question: TR-606, TR-808, CR-8000 all released in 1980 but the 808 used a different selected transistor, why? Why didn't they continue the 808 production with the selected 945p transistors in abundance that fed other machines?