CR-x000 snare

Nevermind 80’8, what about 81’82?

Overlapping the TR-808’s 80-82 production run the CR-5000 and CR-8000 were made by Roland from 1981 to 1982. They both feature a dual bridged-T snare drum circuit, practically identical to their big brother. You can find those who claim that these drum machines have identical sounds, but, as I’ll explain, it’s not that simple. In common with the 808, the SD tuning was updated during it's production life.

The CR x000 Service Notes are headed First Edition December 8, 1981 (reprinted July 12 1983). That’s six months after the updated 808 notes, so not the same moment. The CR-x000 design must have been in the final stages by mid-1981 though, the same period as the 808 tuning changes, right? No! Wait! We can do better than that. Much better. 

 

All Change! All Change!

Flick through the notes to the voice board pages till lo and behold: CHANGES IN COMPONENT WITH SERIAL NUMBERS

CR-5000 101400

CR-8000 101300

 

 

The changes are all over the board but the biggest differences are in the SD!

 

I'm excited. Are you? 

 

Well, before we get to those changes, the quoted serial numbers can be dated to September 1981. And there are other serial number notes. 11700 from October '81 and 091100 from August. This is all after the May '81 changes on the 808 and seemingly after production started in the CR-x000s. So, what changed and to what effect? 

 

Same Old Tune

Firtsly, there is a very comprehensive table of values for the frequency, amplitude and decay time of all the voices. This is similar to the infamous (well, to me) "values are typical and variable" table in the 808 service notes. For the CR-x000 though we get a mininum, typical and maximum value for all three parameters. It's also labelled with the same September '81 serial numbers as above. This means that these values are the revised ones - as you expect, given the December date of the notes. 

 

 

 

Umm, did I say table? I meant to say tables. There are actually four such tables in the document, although it appears that the second page is almost identical to the first and for SD tuning frequency at least I could see no differences. The second table is the values from before the component changes and I took the below image from the second page as it clearly states that these are the prior serial numbers.

For clarity, here's my own table showing how the SD tuning frequencies compare with the original tuning first and the revised second. 

S/Ns	High/Low	Typ
Up-to 101399	H	263
	L	227
101400 onwards	H	333
	L	227

 

 

As you can see, the revision simply changes the ratio between the two oscillator frequency, from 1.15:1 to 1.45:1. and pitches the snare up overall. This is going the opposite way to the 808 which was pitched down. Hmm. But how does that compare to the 808?

Hey, let's just graph the progression from late 1980 to late 1981 and see what the Roland team were doing with their snare drum tuning. 

I'm using the calculated values for the 808 revsions and the typical values for the CR-x000.

Anyway, they've been on what we'd now call a journey there, haven't they? They started off married to the idea of a 2:1 ratio and then weakened that a bit on the 808 revision. For the CR-x000 they almost give up on that idea completely and initially at least have almost identical values. So close in fact that it looks like they made a mistake. Surely not?

In any case they changed it so that the final CR-x000s were pretty close to the revised 808.

In fact the oscillator at the top of CR-x000 schematic ends up almost identical to the 808 on the right of it's schematic. They both use the same component values except for one of the capacitors. On the 808 revisions they only changed one or the two bridged-T caps. 

Go figure. 

Actually, this is where is does get rather confusing. Yes, here. Are you keeping up?

The CR-x000 changes flip the high/low pitch of the two circuits. What was the high (263Hz) becomes the low (227Hz) and vice-versa. So, although it looks in the table like only the high oscillator changed, both sets of components were changed and they swapped places. Even more befuddling is that the very similar valued 808 oscillator ends up with a frequency very close to the CR-x000 oscillator which has totally diffent component values, not the one with very very similar ones! So watch out for that when comparing them because it threw me off track for quite a while.

That all said, the results on the graph above tell the story of Roland's similar but not identical snare drum sound across four different instrument interrations.

 

 Decaying

 

But what about those decay values? 

 

The changes in component values alters the decay time from 64/40ms (H/L) to just 10/34ms (H/L). Given the swap arounds in circuit used for H/L those changes aren't quite as drastic as it might first seem, but they are still 30ms shorter and getting on for a half the length. Not a surprise, given the increase in pitch but still a drop overall all given the lower osc is.. wait, what? The original H osc is longer than its' low osc? Hmm! That calls for a calculation check!

 

Well, the calculation makes it even more stark. The calculated values are 76/42 going to 7/35. after rounding. On the 808 16/51 revised to 22/69 with 60ms as the single typical decay period . Taking these into account, the 808 and CR-x000s look less similar than ever and if you think they sound a bit different, you're not wrong!

At this point I should refer you to the min and max values in the tables. I haven't done any calcaultions on these but I imagine these are accounted for by component tolerances.

One thing I noticed, apart from the quite wide varations, of which more in a moment, is that on the original you could theoretically end up with two indentical osc frequencies more easily. The minimum high osc is the same value as the typical low osc. Whether that has any bearing on the changes I can't say. 

The range of tunings possible are wide, though this may be misleading. The vartiations on components would have to all be in the same direction and at their worst to get to those values. In fact the total population of instruments would more tightly cluster themselves around the typical values and on average you can say that the two instruments are really different and unlikley to ever sound the exact same.

 

Conclusion

 

After all that, what makes no sense to me is that:

1. They seem to have brought the CR-x000 closer into line with the TR-808 after first trying a very similar pair of frequencies, when the 808 was already optimised.
   
2. They altered both oscillators when they only seemed to need to change one, to get the desired frequencies. 

My theory on all this is that ultimately you want all the drum machine voices to sound good together. The CR-x000 BD, for example, is nothing like the 808 kick. For some reason they went with two oscillators and none of the fancy enevlope of the frequncy to get that punchy leading click. On the other hand, the cowbell is slightly different tuning, but only from 540Hz to 555Hz on one of the two oscillators. They just had to be different.

Now, given what I have learned about the CR-5000/8000 SD sounds, can I say that my 80'81 design will also mimic those machines?

Firstly, the decays will never match. As noted above the component values are too different for that. There are areas of overlap in the frequencies, although the way I intend to make the 80'81 works against this. With a single switch for going between the different capacitor values there is no way to have the high osc on a different revision to the low osc. Here's a slightly confusing table trying to make that clear (possibly).

So, you see, that the switch position has to be in the revised 808 position for the only possible way to get the revised high osc value for the CR-x000 and in the mid or original positions to get the low osc values in range. You can get very close to the revised CR-x000 values in the mid position though. Close enough that I don't think it's work going to the bother of a second switch. And as you can see, I hope, the ranges I have do not cover the original CR-x000 high osc at all.

The other thing that makes this CR-x000 option difficult to add in, and the 808 revision much simpler, is that the 808 only changed one out of each pair of capacitors. The switch can thus be a two-gang three-way, not a four-gang or need two switches. 

A single switch is so much better in design terms that this conclusion is a relief!

I haven’t mentioned the noise, or snappy, part of the CR-x000 SD circuit. It is not as sophisticated as the 808’s swing VCA based envelope. Instead a much simpler envelope shaper is employed and the metallic distortion the 808’s design is missing from the CR-x000. This is of course another reason that the superficialy similar sounds are not quite the same. 

 

In summary then: they mnight share some DNA but the CR-5000 and CR-8000 are cousins of the TR-808 and not sound-a-likes.