Roland CR-8000 Modifications

Little jam featuring modifications of clap, bass drum and snare drum.
Cr-8000 in its yet unaltered state

General info and resources:

Although in 2020, prices for the CR-8000 are mostly beyond bonkers, chances are still higher to find an affordable one, especially in comparison to the so-called rich person’s CR-8K, Roland’s own TR808. And while its sequencer design and lack of even basic tone controls for individual instruments may make it seem more desirable for collectors than techno- or hip-hop-heads, it can be modified with relative ease to give you a broad sonic palette and some respectable low-frequencies too.  

Semi-Automatic, who also implemented an elby designs trigger interface, has good general info on the CR-8000 and its circuitry. Many of the drum instruments of the CR-8K are close to those in the TR808, especially the clap, the cymbal and hi-hat section, and the clave. The clap utilizes a “rattling” component and some filtered white-noise. The cymbal/hat section utilizes the output of six square-wave oscillators (produced by the Schmitt-Triggers in a Hexinverter IC) as its source sound, which is the filtered and sent through simple VCA’s. Sounds like the kick, snare, toms, congas, and clave, in turn, are based on so-called t-bridge networks, which are highly resonant filters that are excited by a trigger spike. For snare and toms, you have a bit of filter noise in addition. While the basic building blocks of bass drum and snare are the same as on the 808, there are pronounced differences in how they are set up. While the 808 kick has sophisticated auxiliary circuitry around its single t-bridge network (a pitch envelope, a feedback buffer for decay, and a variable low-pass filter for toning the high click on the note onset down), the CR-8K kick does without envelope but has two t-bridges. As for the snare, on the CR-8000 you cannot tone-blend between the two t-bridges as you can on the 808, and the snappy dial, with which you can change the volume of the noise component of the sound, is missing too. Most of these differences can be altered easily, however. The rim shot on the CR-8K, finally, is unique in that it is ring-modulating the output of the two lowest Schmitt-Triggers that also feed into the cymbal/hat circuitry.

As for the Accent circuit, which allows for dynamic volume control of individual instruments, the CR-8K shows a real shortcoming in my opinion, since not all instruments are affected by it, and since it also merely achieved a level boost of the individual instrument signals rather than boosting the trigger of the instruments themselves (again, as on the 808), which would also result in timbral changes. (This being a reason for me to re-route the instruments and harvest the BA662 IC used in the accent for different purposes). Luckily, individual analogue trigger ins, which respond to velocity, can be done easily, and there are several retrofit midi solutions for triggering too.

The grouping of instrument level controls is also maybe a bit uncommon: bass drum has a volume pot of its own, SN/RS and Toms share one, congas have one of their own, as does the clap, claves/cowbell are shared, and cymbal/hats too. This calls for tinkering single instrument outputs.

The three PCBs of the CR-8000.
CR-8000 voice board PCB left side
CR-8000 voice board PCB right side

Service Manual Scans

Making Instrument and Accent Trigger Inputs

Note that the CR-8000 uses negative trigger pulses. While +5V are present on the instrument trigger input when sound remains silent, a negative-going spike (to 0V) sent from the CPU triggers the instrument. Probing around with this and my Expert Sleeper’s ES3, I found that while the hats, cymbal, and cowbell also take positive triggers without audible difference, the other instruments are sensitive to polarity as well as the length of your trigger – this makes adjustment, especially for the bass drum a bit fiddly. Since sending 0V to the ribbon connector feeding the triggers from the CPU board to the Voice board might harm the CPU, however, you cannot solder your triggers directly there (hence the points elaborated by tubbutec and others; see links below).

If you can do without the internal sequencer like me, simply use the large PCB connector from the voice board to the CPU board, which is the neatest way of setting trigger inputs for all instruments (and accent) up. Here too, all triggers need to be negative (with the exception of the hats/cymbals) but trigger times are not as crucial. Triggers from the CR-8K CPU itself vary between 4ms and 70ms, depending on sequence-tempo, and I found any trigger time 1ms and 100ms from an external source sounded the same.

if you can set your sequencer to send negative triggers, all fine, but if not, here’s a little circuit to convert them to what the CR8K needs.

Pos TRG/GT to negative trg converter

The schematic above is for two instruments only, so you have to repeat the diode-transistor thing per additional instrument. The circuit itself is a simple gate to trigger converter which, pulls your CR8K trigger in (on the PCB connector) to zero volts as soon as there is positive voltage on your EXT TRG input connected to the NPN transistor base. This works well with trigger and gate signals. NB this setup is not velocity sensitive, but works as it were triggered by the internal sequencer.


And here’s one in replacing those tact switches:

CV Inputs for some Instrument VCAs

Several instruments in the CR-8000 comprise VCA circuits, which can be controlled by external CV. Especially for the cowbell and the cymbal section, this allows for experiments and variety beyond simple triggers. Sending an LFO or an envelope is as much fun as manual control. For the hats/cymbal, this could also save you a decay control modification, since you can just send out an envelope from CV-generating device. VCA CV voltages are all positive, except for Clave, and tested up to +5V. For circuit protection, your CVs should go through a diode (cathode to CR-8K circuit) and a small-ish resistor.

Hi hatanode C85/ bottom R153
Cymbalanode D39
Cowbelltop R49
Rim shotanode D49
Clavetop R113 (negative polarity only!)
VCA CV Input modifications for CR-8000 – top/bottom directions per PCB labelling (volume knobs facing down)
CR-8000 hat VCA fed by CV decay envelope and LFO
CR-8000 cowbell VCA fed by sample-hold CV
CR-8000 rim shot VCA fed by CV
CR-8000 Clave VCA fed by CV (features also pitch modification)


A most valuable source for instrument modification of the CR-8000 are Gumi Electronic’s pages with detailed info: Although I deviate from some of them quite abit, they are a go-to source and A++ circuit map!

Bass Drum

As is, the tone of the CR-8000 bass drum is nice and sits well in the sound spectrum, yet with little effort it can also sound way heftier by tweaking the fixed sound parameters. In particular tuning and bass drum decay are rewarding, but I also find changing the volume ratio of both t-bridge networks beneficial. These three mods might not get you to the very final per cents an 808 kick, but they do make the room shake!

Bass Drum Decay 808-style:

With no decay control on the kick, Gumi Electronics calls decay as a have-to-mod. While their suggested alteration offers an extension of the decay time for the lower T-bridge only, I wanted something more flexible. Even though I ultimately went with the “alternative version” described below, a nice mod with which you can adjust decay for both T-bridges with one pot is this adaptation of the feedback circuit in the TR808.

CR-8K kick with lowered tuning and extended decay – hear the falling/”sighing” nicely
CR-8000 bass drum decay modification

In the 808 circuits, one half of IC12 (808 schematic) serves as a feedback buffer, sending the output of the T-bridge network back into the latter. When at extreme values, this lets the T-bridge self-oscillate, thus increasing decay time. One can copy this behaviour on the CR-8K by tapping the output of the buffer amp (IC1, CR-8K schematic). The suggested mod does not require you to take out any components or cut traces. If desired, you can also implement an ON/OFF switch (connection between VR1A and original circuit).

In this schematic, the sum resistance as set by R1A+R2A and R1A+R3A determined how much output signal is fed back into the t-bridge networks. If values are too low, the circuit starts self-oscillating, so values for R2A and R3A are chosen that when R1A is turned fully clockwise, both t-bridges are just short of self-oscillation. (NB that when you also have the BD tuning mod installed, that lower pitches will lead sooner to self-oscillation; also, changing the output volume each T-bridge as described below will also affect feedback/decay behaviour).

If you intend to use your BD for kick drum duties only, it makes sense to have the decay of the higher t-bridge shorter (i.e. have a higher value for R3A) in relation to the lower t-bridge since this augments the impression of a falling pitch “sigh” and leads less to a muddy sound.

Independent decay settings per T-Bridge: If you want to take this circuit beyond conventional bass drum duties it makes sense to use two individual 1megaohm potentiometers in place of VR1A, which comes especially handy, when also using variable tuning modifications for both t-bridges.

Bass Drum Tuning

Again, Gumi Electronics suggest a great mod for tuning the bass drum, which I just modified slightly. I wanted a lower bottom frequency for the kick, so I replaced R24 and R27 (both 680K) with 1.5M resistors each first.

For variable pitch of the lower t-bridge, connect a 1k-2k7 resistor to ground. The other terminal of your resistor goes to a 100K pot, the wiper of which goes to the inward facing terminal of R1 (junction R1-C14-C19). For variable pitch tuning of the upper t-bridge, wire an identical setup to the inward terminal of R7 (junction R7-C12-C13). The range you obtain from this is reaches from very low kick to a conga.

Change Volume Ratio of T-Bridge Networks

On the CR-8K, the lower-sounding T-bridge network louder than the higher-tuned, yet if you want still more oomph, wire a 100k resistor across R30 (68K – the schematic is hardly legible here).

Change between low/igher level of T-Bridge 1 of bass drum
Level went up a bit too much – not enough for gabber though

Using the BD as a Resonant Filter

When working on the RD8, I was amazed to see that there is practically no info on mooders using the t-networks in the 808 or comparable drum machines as the crazy filters they are. Especially when using variable tuning it’s great fun to feed something other than a trigger spike into them and see what happens. The range goes from drone machine of doom to modal resonator pluck. Here’s an example of drumloops fed into the CR-8K bass drum:

External audio fed into the t-bridges of the bass drum

With standard tuning this method is not as fruitful as when feeding audio into the tom and conga circuits, but when tuning and decay mods are installed too, there is oodles of fun to be had.

Snare Drum

CR-8K snare circuit

As mentioned the snare of the CR-800, just like the bass drum, are tuned to be on the tamer side. This can be changed, however. Firstly, being able to determine the volume ratio between the two t-bridges is as useful as tuning them. Secondly, volume control for the “snappy” portion, i.e. the filtered white noise component brings very fruitful results.

Gumi, and have some good documentation of possible mods (NB that 9bit refers to t-bridge 2 as “snappy”).

Extra Snare Output with Extended Volume Controls

The snare sound consists of three components: two t-networks (lower plopp, higher plopp) and a white noise burst called “snappy”. The outputs of these three are summed with R32, R33, and R34 together with the rim shot and the toms at IC 1-A. You can easily change the volume-ratios of the individual sound components, for instance, make snappy louder, by changing the values of R32-34.

For yet more tone-shaping flexibility, however, it may make sense to use an additional op-amp half for dedicated snare mixing, which can also be helpful for having a separate snare output with higher volume. For this, I use an additional LM4558 which piggy-backs onto IC 1 (PIN8 to +15V, PIN4 to -15V). My snappy (junction R95-R34) goes to a 10k resistor. T-bridge 1 (junction R96-R33) goes to a 20k resistor, T-bridge 2 (junction R78-R32) goes to a 15k resistor. All three resistors go to the –IN of half a LM4558. –IN and OUT of the op-amp are connected with a 100k resistor, +IN goes to ground, and OUT goes via a 1uf cap (anode to OUT) to a separate jack socket. IUf you want variable volume control of this mix, use a 100k pot instead of a fixed resistor between -IN and OUT.

Things start to become more interesting when you implement options for tone-blending. Inspired by the 808, I have both resistors of the t-bridge outputs connected to the outer pins of a 10k pot and the wiper of that to the -IN of my op-amp. This way you can blend between lower and higher t-bridge. You can have the output of your snappy/noise component on a pot as well, yet I find the following 808-style snappy dial more rewarding.

Toying around with the CR-8K snare (first 4 hits unmodded)

808-Style Snappy Level Control

As far as I understand the 808 snare snappy control, this is more than just a simple level dial since it does not just regulate the level of the circuit’s output but rather how much of the trigger spike is fed into the amp envelope circuit (resulting in different cap charges/discharges, i.e. quieter and shorter versus longer and louder snappy). In order to halfway transfer this to the CR8000, lift the cathode of D28 from the junction R100-C49 and insert a pot with a value between 250k-1M. The exact value depends on your general WN level. Without WN input boost (described below), I get good results with a 33k resistor and a 250k pot in series (which go between the free cathode of D28 and the junction R100-C49). With WN boost, I use a 500k pot but still have a tad of snap (fine with me) when the pot is fully closed. If you use this and the extra snare out, a separate level control pot for snappy is not required.

808-style snappy mod fed with trigger of different velocity – note how louder notes also sound longer
Snappy bit of CR-8000 schematics

White Noise Input Boost

Even if you raise the output level of the snappy portion by lowering the value of R34, there is still something missing in terms of a sharp snap. So if you want to take this snare a bit more to the aggressive side, increasing the input level of the white noise entering the snappy circuit does the trick! Lower the value of R123 (100K) either by replacing it by a resistor and pot in series or wiring a resistor in parallel to R123. Exact values depend on your machine (especially the setting of the white noise level trimmer, VR11) and taste. There is a sweet spot, when the noise just starts to really crunch but not “hiss” (for lack of better words), and on my unit this is with 20k in parallel to R123.

Louder white Noise input CR-8K Snare

Alternative Noise Input for Snappy

White Noise enters the snappy circuit at C52 (0.1uf). If you want some alternative inpiut material, say the metal noise from the cymbal section, just lift the terminal of C52 connected to the junction R122-123 and wire it to the normal pin of a jack socket. If you plug another signal in, the connection to white noise breaks and your alternative signal is fed in instead. Schnappay, folks!

Alternative noise input for CR-8K snare

Longer T-Bridge Decay

While altering R100 (1M) changes the decay time of the snappy portion, the t-bridges themselves need a trick similar to the BD decay extension as described above. I think it’s worthwhile to increase the t-network decays a little bit, but this gets easily too much and maybe they are just fine as they are anyways 🙂

Be it as it may, I connected the output of my additional summing op-amp via a 250k resistor to the junction R79-C38-C39. This extends the decay of the lower t-bridge but also shifts its pitch a little bit up, so I lift the ground terminal of R79 (15k) and have a 100k pot between R70 and ground, which gives a nice range. NB that when you extend the decay of both t-networks this way, you get also pitch interaction ebtween the two. While this happens with the BD decay mod too it is less noticeable there but more so on the snare. For proper independent decay, it might make sense to use two independent buffer amps.

Lower t-bridge longer on snare
loinger decay on bth t-networks of snare (listen how both pitches/decays show interference at extremer settings)

Tuning the T-Networks

Tune the lower t-bridge by changing the value of R79 and the higher t-bridge by changing R94. Gumi and 9bit have detailed infor on this.

Tom Modifications

As they are, the toms sound nice but lack some impact in my opinion, so I firstly reduced their output attenuation and also rewired them to the clave mixer section (IC8, PIN6) in order to have snare & rim on a “channel” of their own.

In terms of sound mods, variable tuning is definitely useful, and more extreme mods would be T-Bridge warble (i.e. feeding CV/audio into the T-Networks, using them as filters/resonators), and som einterdependent overdrive (“Tom Mayhem” described below).

LT tune: replace R8 (68ohm) with 500ohm pot [If you use one of the “Tom Mayhem” mods described below, a better range can be had by using a 1k pot with a 4k7 resistor across the pot pins. Without distortion, the tuning would be too low to have some impact but introducing “mayhem” drive takes quite some overtones up, so lowering the tune range yields cool  effects.]

HT tune: Replace R43 (68ohm) with 500ohm pot

CR-8K variable tom tuning (also with increased output level)

LT level: replace R37 with 10k resistor

HT level: replace R38 with 10k resistor

LT audio/CV input (adapted from RD): right terminal R9 via 1k (for impressions see “BD as filter example above”)

HT audio/CV input (adapted from RD): R43 via 1k

Tom Noise Level: replace R35 (33k) with 50ohm/100ohm resistor.

Tom Mayhem: With this modification, you basically feed the output of the LT T-Bridge network either into itself or into that of the HT, which introduces overdrive and also some mutual pitch interference/co-triggering. In combination with variable tuning, this mod takes you quickly to techno turf!

Wire the right hand terminal of R26 to a 1k resistor, which goes to a 25k pot, which, in turn, goes to a rotary switch. Destination A of that switch remains unconnected. Destination B goes to the lower terminal of R44 (which is part of the High Tom T-Bridge). Destination C goes to the lower terminal of R24 (part of Low Tom T-Bridge). Finally, destination D of your switch goes via a 10k resistor to the upper terminal of R44.

What it sound like is roughly described (also listen to demos below):

A = no mod

B = punchy overdriven HT

C = punchy overdriven LT

D = grimy overdriven LT

That’s what those “vanilla” CR8K toms can sound like (some delay fx added)
Another example of cross-overdriven toms

Proper Volume Control (this is speculation ATM, i.e. needs testing): Since LT and HT share the same noise circuit, having volume control per tom is less easy than just replacing a resistor with a dual gang pot (this way, having zero volume on one tom would also silence the noise component of the other tom).

So what you need to do is to use a dual gang pot one part of which (A) attenuates the T-Network output level and the other part of which (B) attenuates the trigger input voltage for the noise component (say, if you do this for LT, then triggering HT would still trigger the noise as it should). Since I have not tested this yet, I take 100k as a hypothetical pot value.

LT volume control: pot (A = T-Network): R39 replaced by pot and 10k resistor in series with third pin to ground; pot (B = Noise): lift cathode D19 and insert pot, with third pin to +5V (remember, triggers work with negative polarity in the CR8K, with +5V present on not-trigger state)

HT volume control: pot (A = T-Network): R38 replaced by pot and 10k resistor in series with third pin to ground; pot (B = Noise): lift cathode D20 and insert pot, with third pin to +5V