We are here: [ home ] -> roland_stringschoruses

Roland Choruses and Ensemble Effects

This is work in progress and shall become an overview of chorusses and ensemble effects from Roland in the pre-digital era. Maybe I will extend it to other brands too.

The differences between various choruses are caused by several parameters: basic delaytime of the BBD, modulation rate, modulation amount. And this still neglects the influence of filtering and dynamic noisereductions. At the moment I can provide only the modulation rates. To provide the basic delaytime and the modulation amount it is required to modify the circuits and measure inside the running machine - this might take some time...

String Ensemble Choruses

String Ensemble Choruses are a quite simple method to imitate the sound of orchestral string instrument ensembles. Since the 18th century string ensembles played held notes always with vibrato. Since the vibrato of each player was different in frequency and phase, this resulted in a very rich and warm tone. The ensemble chorus uses the simple fact, that a change in the playback speed of a delay line causes a pitch change of the delayed signal. If the delay line is modulated by a triangle wave, then a kind of vibrato is achieved(*). If a straight signal without any vibrato is sent into several parallel delay lines with different modulations, then the result sounds like several signals with different vibratos - vulgo like a string ensemble. Most string ensemble keyboards of other brands used three delay lines with a single modulation signal, which was assigned with a phase shift of 120 degrees each to the three delay lines. Roland used this only in the RS-202. All other Roland string synths were based on a mix of the stereo-chorus principle (see below the synth section) and the ensemble effect of several different modulations.

(*) In fact it is not a vibrato but a trill: the down slope of the triangle makes the pitch constantly lower, the upward slope makes it constantly higher. As long as the modulation amount is not too high, no one will hear the difference between vibrato and trill
Instrumen Effect Type Topology / Data Comment Schematic
RS-09 Ensemble BBDs: 4x SAD512D(*)
4 LFOs

LFO1 (0,15Hz)
LFO2 (4,7Hz)
LFO3 (0,18Hz)
LFO4 (5,9Hz)


BBD1 + BBD2 -> Left
BBD3 + BBD4 -> Right
There is an external input socket, which allows to send any signal into the string-chorus.
The original signal is disabled, if the ensemble effect is enabled.

(*)This is from the service manual RS-09 MK1.
I assume, that the RS-09 MKII is using MN3004.
RS-101 Ensemble BBDs: 4x MN3002

1 LFO divided three times
Base Frequency 12Hz
Divided 6Hz, 3Hz 1.5Hz
There are two interesting aspects:
* the 1:2:4:8-Ratio of the LFO rates,
* and the simulation of the LFO itself. They simply seem to use the rectangle from the divider chip and "integrate" it by loading / unloading an electrolytic capacitor. This will look similar to an triangle (in reality more like an AD-envelope shape).

Unfortunately I don't own an RS-101, so I can't prove my assumption.
RS-202 Ensemble BBDs: 3x MN3002

6 LFOs organized as three pairs of LFOs, in each pair one LFO runs at 6.25Hz the other LFO runs at 0.66 Hz. There is a "Chorus  fast" switch, which disables the slow LFO.
Interesting aspect: Roland did use three separate LFOs, which trigger each other to provide a 120degree phase ratio. This looks like much more effort than using a single LFO with phase shifted outputs. But assumingly it was cheaper to use a lot of transistors instead of OpAmps. Blockdiagram
RS-505 Ensemble BBDs: 4x SAD512D
4 LFOs

LFO1 Triangle 0.17Hz
LFO2 Sine 6.25Hz
LFO3 Triangle 0,14Hz
LFO4 Sine 5.5Hz


BBD1 + BBD2 -> Left
BBD3 + BBD4 -> Right
After the more Solina-like RS-202 Roland went back to the configuration which they used already in the RS-101.
Also here we find an external input for the chorus and a dry-signal out socket. Unfortunately those two do not form an insert. If it would be an insert, one might extend the RS-505 with an external vocoder to a poor mans VP-330.
VP-330 Ensemble / Vibrato BBDs: 2x MN3009, 4x MN3004
5 LFOs

For LFO 1 to 4 the service manual does not mention any frequency values. I assume they are quite similar to the RS-505.
LFOs 1 - 4 go to the four MN3004 in the same way as at the RS-101 and RS-505.

LFO5 is free adjustable in frequency and modulation amount. It controls the clock of the Two MN3009.
Basically the ensemble effect  in the VP-330 is very similar to the RS-505. Some people out there mentioned, that possibly the VP-330 MkI used SAD512D - at least for the string chorus. This is quite likely, as the MkI was released at the same time as the RS-505 (which uses SAD-chips). But the 2nd-edition service manual does not mention a change in the BBD. Unless I don't find a picture or an older schematic, this topic ios not proven.
The more interesting part is the  usage of the two MN3009:  They are intended to provide a vibrato to the "human voice" and the "vocoder" section. Roland might have simply added an LFO to the top octave oscillator to achive vibrato (as they did in the RS-505), but then the vibrato would have left the voice-simulating filters static. Using the BBD for the  vibrato achieves that also the frequencies of the filters (vocoder bands and male/female filters) swing with the vibrato.

Mono to Stereo Chorus

The famous "Roland Chorus" found in Rolands middle class polysynths is based on a simple principle: two delay lines are modulated by the same LFO, but the modulation is inverted for one of the delay lines. The two delayed signals are sent to the left and right output, the dry signal is mixed 1:1 to both delaylines. This construction outwits our hearing: we do no longer hear the pitch modulation which happens due to the modulation of the delay.
Jupiter-4 Mono to Stereo Chorus BBDs: 2x MN3004

1 LFO with fixed frequency - assumingly around 0.7Hz (roughly estimated according to the schematics; has still to be measured)
SA-09 Mono-to-Stereo-Chorus,
switchable to Mono-Vibrato
BBDs: 2x MN3009


Chorusmode 0,24Hz
Vibratomode 3,8Hz - 7,9Hz
In the chorus mode one of the BBD-paths is frequency split: Lowpass->BBD, Highpass->Original. This reminds the topology of the Dimension-D.
The other BBD path covers the full frequency range!
Juno-6 Mono to Stereo Chorus BBDs: 2x MN3009

LFOs: 1LFO with three settings
* I = 100% amount, 0.4Hz, triangle
* II = 100% amount, 0.6Hz, triangle
* I + II = 25% amount, 8Hz, sine like
The complete chorus and LFO section is on the right panelboard. In the "fast mode" the second BBD is controlled in a non inverted way.
Marc Renton made a very nice video about this Chorus
Juno-60 Mono to Stereo Chorus BBDs: 2x MN3009

LFOs: 1LFO with three settings
* I = 100% amount, 0.5Hz, triangle
* II = 100% amount, 0.8Hz, triangle
* I + II = 8% amount, 1Hz, sine like
The chorus circuit with the BBDs itself is placed on a separate "Chorus Board". The LFO is placed on the right panelboard. The (programmable!) different speed settings are achieved using VC-switches. In the "fast mode" the second BBD is controlled in a non inverted way.
Marc Renton made a very nice video about this chorus
Schematic chorus
Schematic LFO
Juno-106 / MKS-70 Mono to Stereo Chorus BBDs: 2x MN3009

LFOs: 1LFO with two speed settings
In the service manual there are no frequencies defined for the chorus LFO, but I assume they are similar to the Juno-60.
In opposite to the Juno-6 and Juno-60 there is no longer the possibility to set the mode "I+II" to get a higher chorus speed. Also the deactivation of the inversion is no longer available.
Marc Renton made a very nice video about this chorus
Alpha-Juno-1 / MKS-50 Mono to Stereo Chorus BBDs: 2x MN3009
the modulation voltage is generated in the digital domain, the speed can be set in 127 values. There is no hint about the available frequencies.
The Juno-1 is the first Roland synths that used compression/expansion for noise reduction in the chorus section. The compression is not achieved by the usual NE570/NE571 pair. Instead of Roland used a "compressed" envelope voltage for the VCAs and then placed a second VCA after the after the Chorus lines, which was fed with an "expanding" CV. Schematic
Alpha-Juno-2 Mono to Stereo Chorus See Juno-1 Same as the Juno-1, but in the Juno-2 the Chorus is no longer on the mainboard, but on a separate chorus board see Juno-1
JX-3P / MKS-30 Mono to Stereo Chorus BBDs: 2x MN3009
1 LFO with fixed frequency at 0.25 Hz.
The MKS-30 has slightly different part values at the LFO, it is possible, that the LFO speed is slightly faster (0.35Hz?). The service manual does not mention the frequency. There is a modification from Kiwitechnics, which replaces the analog LFO by a CPU control for variable speed and manual delay time setting.
Marc Renton made a very nice video about the JX-3P chorus
JX-8P / JX-10 Mono to Stereo Chorus BBDs: 2x MN3009
1 LFO with two fixed frequencies: unfortunately the service manual does not mention the frequencies.
The LFO speed setting is similar to the Juno-106, but not the same. Interesting aspect: there is a very simple limiter circuit in the audio-in of the chorus. Appearently the roland engineers found that the JX-3P can overdrive its chorus; assumingly they wanted to avoid that without making the sound/noise ratio worse. This limiter is not(!) in the signal path of the dry signal, so it might be that percussive sounds are more "incisive" at the JX-8P than on the 3P. Schematic
MKS-20 Mono to Stereo Chorus BBDs: 2x MN3007
1LFO with programmable "intensity". The intensity is a combination of modulation rate and modulation amount: the faster the modulation, the lower is the modulation amount. For the values see MKS-20_modulation.jpg
The chorus in the MKS-20 is interesting under two aspects: 1.) Its basic layout reminds the Dimension D. It mixes the BBD-signal of one channel also with 30% to the other channel. This avoids that the trill effect is too much appearent, which might happen if the chorus speed is quite fast. Also it uses a compander circuit for noise reduction. Whilest they used already MN3007s, which run at a much higher clock speed for chorus, which again improves the signal to noise ratio.
2.) The LFO can be programmed in speed and modulation amount. Roland did not use a VCO for the speed, and not a VCA for the amount, though that would have been easy. Instead of they used digital switches to switch between various resistor values for the LFO and the buffer amp.
Adam Inglis provides an instruction for an external input modification to the Chorus: http://www.adambaby.com/studiotech_MKS20.html

LFO values for amount and frequency

Dimension-C to be done


MKS-10 Mono chorus / vibrato BBD: MN3007
1 LFO: amount and depth are variable, but not programmable
The chorus in the MKS-10 is applied to one side of the stereo-output only. This way it reminds the "stereo"-concept of the JC-120/JC-160 Blockdiagram
LFO schematic
Chorus schematic
SE-301/SE-501 to be done
JC-120/JC-160 to be done


SBF-325 Stereo-Chorus / Flanger 2 x BBD MN3007
The SBF-325 can be seen as the completely editable version of all Roland chorus'es and flangers - at least in respect of the parameters. The effects mode "Flanger III" is the same as the Dimension-Ds algorithm. So if you leave away the feedback and choose the right delay and modulation settings, it sounds like the Dimension-D. Blockdiagram
SDD-320 Dimension-D Stereo-Chorus 2 x BBD MN3007
The Dimension D is a kind of special chorus device. In the opposite to all studio flangers and choruses it does not provide any direct access to the effects parameters. There are only four buttons, which select between four presets, which are called "modes". Also it works with a kind of unique algorithm. The signal runs through a passive 3dB Highpass around ~80Hz, then is fed into the modulated delay stage, which uses a compander circuit for noise reduction, and finally the modulated signal of each side is mixed with its own original signal and with the original signal of the other channel. In "mode 4" the mix to its own original signal is padded by -10dB. The parameters of the four modes are like this:

When I announced this list on the analog heaven maillist, several people made useful comments and contributions. I will collect those here too:

Replacing BBDs for noisereduction

In general BBDs detoriate when they are getting older. After 30 years they can become noisy and crackeling, and they may start to distort the audiosignal. I could improve the sound of my Juno 6 and of a Hohner String Melody by replacing the BBDs with new old stock chips.
Murray Hodge from kiwitechnics.com contributed the following "A tech friend of mine tried replacing the MN3009 chips in a 106 with the 4x larger capacity MN3007s which can still be found new and have the same pinout. He then raised the clocking speed 4x by changing the Resistors 124 & 137 from 8k2 to 2k2. I haven't tried this myself but he told me the result is much quieter and works well."
Brian Willoughby from audiobanshee sound consulting commented: "Parts are so cheap today that I'd recommend replacing the 8.2 kΩ resistors with 2.05 kΩ resistors. You have to buy 1% resistors, probably metal film, to get that value, but they're only $0.10 in single quantity, so shipping will be more than that anyway. The reason I say this is that the 2.2 kΩ suggestion on that web page could be as much as 13% too high, resulting in a longer delay than original. Assuming the factory resistors are 5%, the 2.2 kΩ resistors could change the tone of the chorus. It's cheap enough to literally divide the original resistor values by 4 and just spend a few pennies more to keep the delay times as close to original as possible. ". Definitely true
I did not try this modification too, but I had a look at the various circuits and found, that Roland always used the same VCO-circuit for the main clock of the BBDs. So here is the list, which resistors have to be changed in which synth: The JX-10, and the Alpha Junos do use a different VCO circuit. As soon as I have more information about their function, I will add the it here.