Rss

Archives for : jv90

The ultimate Roland JV, JD, XV F.A.Q.

jv80

Super JV vs XV series
Following the JV/XP series were Roland’s XV series: 5080, 5050 and 3080. XV-5080 is mixed content 32kHz and 44.1 kHz. I got this later confirmed by Roland. (though some web pages list it as 32kHz ROM only, but this is not true). I will focus now on XV-5050 and compare it with JV-1080. Some users started complaining about the XV-5050 sounding a bit “thin”. There is some truth in that but what i can tell in reply is that 5050 sounds more hi-fi. Because of 44.1k sample content, some energy has been “lost” due to wider frequency coverage. Patches played on 1080 and 5050 side by side will sound different. This is a fact that i’ve verified myself. 5050 is more hi-fi and has that extra sheen while 1080 is more darker and is a bit more mix friendly when it comes to frequency and EQ. You will find some waveforms more hi fi sounding in XV when compared to Super JV series.

It should be worth mentioning that 5050 has some sort of permanent low shelf filter at about 30 Hz, so you’ll definitely get a less bass energy. But the high freq response is just spectacular if compared to something like a JV-1080. Especially when you start using the digital output and route it directly into DAW, it’s a no match in crystal clear sound. FAQ UPDATE according to Joe (from comments below) the 5080 seems to have the same low shelf filter going on like 5050 and they seem to sound identical. This is what i always suspected, however since 5080 can set its clock to 48 kHz when loading S series samples we can’t say they sound 100% identical, simply because 5080 can produce more high freq content in ‘S-760 mode’.

One thing that is very different on 5050 vs 1080 is the dynamics. For some reason it seems that 5050 has some sort of compressor at its output. As a result, some of the patches have less dynamics going on in them. This is most obvious on layered sounds that have a lot of phasing between oscillators going on. While the same patch on 1080 will produce more differences in volume, on 5050 it is more constant. This can be good or bad, depending what kind of sound you need. For movie/TV scores you would probably want more dynamics going on, hence the 1080. And for dance music, you would go 5050 since it delivers that straight – in your face sound – right out of the box, without need to work on dynamics. For the above reasons 1080 definitely sounds more soft and gentle.

As of XV-5080, i tested it side by side against XP-30 on the same patches and the difference was quite noticeable in what appears to be a far greater stereo field and definitely superior sonic quality of 5080 effects. I particularly remember one preset called Letter From Pat. In fact if you have both units, just load it and hear the difference for yourself. It’s day night difference in favor of 5080.

990

JD-990 vs. XV series
XV series contain the whole JD-990 waveform set. With XV-3080 being 32k and XV-5080 and XV-5050 with original 44.1kHz JD set. Some of the waveforms have been renamed, but they are there. It should be said that on along the Adaptive DPCM waveform compression, I always suspected (but never got it 100% confirmed) XV series have extra  (destructive?) form of compression on top, similar to mp3 and it can be spotted visually with most simple analyzer. There is no such compression on JD series. More on that in one of the chapters below.

Patch conversion JD into XV is directly not possible. However it would be possible to convert (manually) a patch from JD-990 into 5050 since Roland implemented the whole “Effects Block A” section from JD into 5050 (available as EFX called JD Mlt). Block B can be emulated with Chorus/Delay and Reverb. There is a whole article on this subject available on this website. Only difference is the filter cutoff numeration system. On JD-990 it goes from 0 to 99 while on 5050 it is 0 to 127.

There were some rumors on various forums that XV-5080 is 32kHz (thus being able to play only up to 16kHz). This however is simply not true. We will now take a look at a waveform spectra of a White Noise sample as played from JD-990 and XV-5080. What we can clearly see is that not only they are identical but they both go all the way up to 22kHz, which clearly indicates 44.1k playback.

02 01

Benefits of XV over JD is that the filter on XV has a greater dynamic range. There is no clipping issue on XV as opposed to JD when you set filter keytracking to 100%, find a resonant spot, press a chord and end up in harsh digital distortion (if resonance is above 40). Not only XV won’t distort, but even if it happens on some waveforms, there is one additional parameter called oscillator Gain that lets you reduce the volume of the waveform prior to being fed into filter. You can set it to 0dB or even -6dB. On JD it appears to be permanently set to +6dB (of XV equivalent) which is a pity. That’s the only feature i can’t regret not having on JD. Of course one thing that is very known is that there is definitely a difference in the high end of the filter. JD-990 will go a little bit higher in frequency and thus add more sweetness. The rest of the frequency range response is almost identical.

1080

The Sound
There has been a lot of talk about difference in sound within units that should be based on the same engine. We will here list the converters used which might indicate why some minor sonic differences. There’s an old rumor that the film guys prefer the sound of 1080 against newer the XV series such as 5050. This is a bit complex matter since it involved dynamics and not just frequency, and i have explained it in a chapter above. Let’s now take a look at converters of JV and JD units (notice: XP is a JV with a keyboard)

JV-80   32k  sample rate DAC: 18-bit PCM69P
JV-90   32k  sample rate DAC: 18-bit PCM69AU-1
JV-880  32k  sample rate DAC: 18-bit PCM69AP (main out)*
JV-1080 32k  sample rate DAC: 18-bit UPD63200GS-E2
JV-2080 32k  sample rate DAC: 18-bit PCM69AU
XP-30   32k  sample rate DAC: 24-bit AK4324
XP-50   32k  sample rate DAC: 18-bit UPD63200GS-E2
XP-60   32k  sample rate DAC: 18-bit PCM69AU
XP-80   32k  sample rate DAC: 18-bit PCM69AU
JD-800  44k1 sample rate DAC: 18-bit PCM61
JD-990  44k1 sample rate DAC: 18-bit PCM61P
* uses UPD6376GS-E2 for sub out
  • JV/XP uses Adaptive DPCM, plus something that looks like a destructive form of wave compression (mp3 style)
  • JD uses Adaptive DPCM and no destructive compression (no data holes)

Some people claim they can hear the difference of JV-1080 vs. JV-2080. Unfortunately i don’t have them side by side to verify this, but if someone can, simply load the same patch, record it and send it to me or on the Gearslutz forum and we will inspect it. The rumor is that 1080 sounds “better”, whatever that means. Only thing i can confirm is that converters on the JD-990 sound way better (more stereo width) than those on JV-1080. In fact, it’s probably the best sounding synthesizer that Roland ever designed. Hearing is believing and you should really give it a try if you didn’t by now. There’s a reason why JD-990 scores for much more than 2080, although from technical standpoint, 2080 offers much more waveforms and has better mod matrix.

Some quick points: Over the years i’ve had following machines JD-990, JV-1000, JV-1080, XP-50, XP-30, XV-5080, XV-5050. From first hand experience: if you want a lot of sounds and not the quality, XP-30 is an absolute winner. It you want max quality, then go either JD-990 or XV-5080. If you care for the high sheen filter sound, go with JD-990 as it can pull out the way XV-5080 can’t. But 5080 has much more waveforms (including some from Vintage Expansion) and has far superior effects, filter dynamic range and modulation engine (it features true matrix system). On top of that it can be used as a sample player since it has a “S-760 mode” (though that limits a lot of synthesis functions).

5080

Compatibility
Even the latest XV-5080 has a full backward compatibility, all the way to the JV-80. You can also load all of the patches from JV-80, JV-90 and JV-1000 into JV-1080 and JV-2080. Just like you can load JV-1080 patches into the last of the series XV-3080, XV-5080 and XV-5050. They are all full compatible with only a few minor exceptions when it comes to waveforms. Even the old JV-80 patch will sound identical if you properly convert it. Some correction in resonance is needed because old models JV-80, JV-90 and JV-1000 had a Soft and Hard resonance setting, next to the resonance amount. Because JV-80 has two resonance settings, Soft and Hard. Their equivalent on Super JV and XV is as following:

  • JV-80 Soft setting, resonance set to max = XV-5080 reso set to 44
  • JV-80 Hard setting, resonance set to max = XV-5080 reso set to 88

What applies to XV-5080 applies to all Super JV and XP series. I came with this info by testing them side by side. This also gives you idea that the filter in JV-1080 can go way beyond old JV in resonance power. This is not surprising since it is a filter from the JD series. To cut the long story short, whenever you load a JV-80 patch into Super JV or XV you will have to modify the resonance value.

Antialiasing filter in Super JV is superior to the one in JV – which, depending on what kind of sound you like, is welcome or not so feature. Mirroring in higher frequencies, particularly when using rich textures can fool the listener thinking the unit is 44kHz waveform set, though in reality it is not, it is 32kHz just like Super JV. I talk about mirroring above 16kHz which can happen during transposition, thought the waveforms are all 32 kHz. This is just an artifact that happens with low interpolation quality algorithms. So in a way, old JV can sound a bit more open than the later Super JV series, because of the weaker anti alias filter in JV.

Patches
JV-1080 contains some of the JV-80 patches. JV-2080 contains all JV-1080 patches plus a bank of additional ones. XV-5080 and 3080 contain all of the JV-2080 patches, plus a few new banks. XV-5050 contains all XV-5080 patches plus a bank of additional Fantom patches (these are located in the User area 1-128).

Destructive compression?
With the Super JV series, on top of the existing Adaptive DPCM compression it seems as if Roland added an extra compression which is destructive form of compression. This is not confirmed anywhere in documentation. But at the same time it is trivial to test that something is going on by using a JD and any JV synthesizers, plus a spectral analyzer. If we play exact same waveform on both, some parts of the spectra are simply erased on the JV/XP/XV version. Now where have we seen that before? The good ole mp3 kinda looks like it, no? Of course it is not mp3 compression, because there was no mp3 back then, but the principle is somewhat very similar. Here is one example that clearly demonstrates it:

upsampled

The same waveform was chosen on JD-990 and XV-5080. Please ignore the mirror effect label on the image, it relates to interpolation and that shouldn’t concern us. If we look at the waveform from 5080 somewhere around 15kHz we can clearly see a hole. There are a lot of such waveforms in Super JV and XV series that have holes in them. Very similar how mp3 works. And as you can see there are no such holes in JD-990 which makes it clear that JD-990 does not have this missing data. JD however use some other form of compression though, but we will discuss that below.

DPCM or a companding compression and Roland
On the Gearslutz forum, in May 2010 Eric Persing (source: here) mentioned that JV-1080 uses 8-bit companding compression. We can assume the same is true for the JD series as well. Unfortunately I can not confirm nor deny this, but I believe the man’s word since he not only designed most of these waveforms but figured out how to actually put them into hardware! What is not entirely clear from his statement was the exact compression method. If it is “phone line companding” type of algorithm – this is relatively old process which goes as following: Once the waveforms are sampled at the factory, they are being dynamically compressed and converted to 8 bit. The reason why they are compressed prior to that is to preserve low level information and somehow increase the dynamic range of this 8 bit file. At that stage they are put into machine’s ROM. Once the machine boots up it will load a waveform, convert it to 16 bit and apply dynamic expansion. Essentially the same thing what a compressor and expander that you have in your rack do, although these have 0 attack / release time. Data compression dates back into days when memory was very expensive, and manufacturers were looking way to squeeze as much as possible into fixed ROM space. Companding was one of the options where for every 16 bits of input, you would use only 8 bit to store them, yet with some tricks “preserve” the data. However, from my own research, and consulting people who have reverse engineered the ROM data of these machine, it seems that Roland does not use companding compression at all. Instead what I believe happened was that Eric used this word to make it more simple for average people to understand, since after all he is constantly in talks with audio engineers, and it would take too long to explain the exact algorithm so he most likely uses this as a short phrase for compression / expansion. The unfortunate bit in here that there was actually a compression method which contained that exact name.

It seems that Roland modules, all up until recently with the 2019 Fantoms, use DPCM compression type which downscales the data into 8 bit by a process of differential pulse-code modulation. This is a signal encoder that uses the baseline of pulse-code modulation (PCM) but adds some functionalities based on the prediction of the samples of the signal in two possible ways: 1) Take the values of two consecutive samples, quantize them, calculate the difference between the first one and the next, the output is the difference. 2) Take the difference relative to the output of a local model of the decoder process and quantize it. Compression ratios on the order of 2 to 4 can be achieved this way.

The question now arises: does that make Super JV and JD series 8-bit machines? Well technically speaking no. These are not just plain 8 bit samples in the ROM but 8-bit compresses samples. It makes a difference, because prior to being played, their dynamic range is restored and expanded to 16 bit. I haven’t meet a person that doesn’t like the sound of Super JV series and they would hardly believe these originate from 8 bit samples – but in a way, they do. In this regard we can also assume when Eric Persing mentioned the “companding” compression he was referring to DPCM, since the data is actually compressed into 8 bit and then later expanded into 16 bit (realtime using dedicated DSP hardware).

Engine and sample rate
Roland JV-1080 has a waveform set which is at 32 kHz. Its DAC runs at 32 kHz. We can see that in the image below. A sine wave was played at 8 kHz, and we can clearly see a mirror effect (aliasing) at 24 kHz. From this we can gather: 24 – 8 = 16. From this, Nyquist on JV-1080 is at 16 kHz. This tells us that a DAC runs at 32 kHz. In fact, just by looking at the picture you can immediately see that the whole image above 16 kHz is “mirrored”. You will have to click on the picture below for full size. Further more, by close inspection we can see a constant carrier wave at 32 kHz which could be the bleed thru signal of the DAC itself. Because i see no other explanation for a constantly preset 32 kHz signal, than the DAC itself.

jv1080

I’ve read on GS forum some people claimed JV-1080 to be 44kHz DAC, but this is simply not true. If it was, then for start, the mirror effect (aliasing) would happen at 22 kHz, not 16 kHz. Another argument given was usually “this DAC can run at 44 kHz”. Yes, that is true. But it can run at 88.2 kHz as well! Even way beyond that without any problem. Looking at chip specs table isn’t always the best source of information. A simple measurement is sometimes all it takes.

Another argument that i read was 32 kHz DAC can not produce frequencies above 16 kHz. If this was true, then the assumption of that same person (original post here) that JV-1080 runs on 44.1 kHz is wrong as well. Because we can clearly see in the image above the unit goes way over 30 kHz. So does that mean DAC runs at 60 kHz? No it does not! The problem in here is the wrong assumption to begin with. A 32 kHz DAC can in fact produce frequencies above 16 kHz. This is considered an artifact and is known as aliasing. Back then manufacturers spent a ton of resources to suppress and remove as much of these as possible. As we can see Roland went for the simpler / cheaper option with some basic LPF filter behind the DAC, far away in specs of today’s brick wall filters. In fact service manual suggest this scenario as well. As a result of all that a lot of signal is aliased.

scope

Image above shows a DAC chip world clock input (pin 13) on JV-1080. Signal is close to 5 volts peak to peak and is running at frequency of 32,00 kHz. The story of JV’s playback and engine sample rate ends here! For those interested in how i’ve obtained the data here’s a full story: In order to verify the assumption about the data shown on spectrogram, which shows mirror at 16 kHz and to be 100% i’ve downloaded specs sheet for the UPD63200. It is a DAC chip which is used in JV-1080. Next step was to find out the pin where the World Clock is located. And that turned out to be pin 13. After that i simply opened JV-1080, and located the chip. Luckily there is a via on the PCB board which can be used to connect the probe to, rather than touching the chip pins and risking of doing the short circuit (thank you Roland). So i connected the oscilloscope probe to pin 13. The result can be seen on the image above. Clock rate of the DAC chip was measured to be exactly 32,00 kHz. Just like we estimated by observing the spectrogram data. This confirms the earlier findings and verifies that JV-1080 is indeed a 32 kHz machine.

History tree

JV89a

Timeline:

  • JV-80 (1991) = a true masterpiece of it’s time.
  • JV-880 (1992) = rack vesion of JV-80.
  • JV-1000 (1993) = JV-80 + MC-50mkII sequencer, added new waveforms, floppy drive, 76 key.
  • JV-90 (1994 ) = JV-1000, without sequencer and floppy.
  • JV-1080 (1994) = huge step forward for Roland. This was the most popular module of 90’s. New filters, voice structures, 448 waveforms, matrix control, new features.
  • XP-50 (1995) = JV-1080 with keyboard, sequencer, floppy
  • JV-2080 (1997) = JV-1080 big LCD (better user interface), 3 EFX, 8 x expansion slots.
  • XP-80 (1996) = XP-50 with 320 x 80 dot LCD (better user interface), 4 aditional sliders, more outputs, arpeggiator, 76 key.
  • XP-60 (1998) = 61 key version of XP-80. It replaced the XP-50.
  • XP-30 (1999) = XP-60 with added patches (waveforms) from three expansion boards (session, orchestral, techno), removed sequencer. By number of factory installed waveforms, this is the most powerfull XP and JV synth!
  • JV-1010 (1999 ) = JV-1080 in half rack module, session patches (waveforms) added.
  • XV-5080 (2000) = another big step forward for Roland. 1083 waveforms, 128 polyphony, true stereo voice – each tone (T1-T4) can be set as stereo (one waveform for the left, one for the right channel), SCSI connection, sample load, up to 128 MB of RAM (SIMM), 5 effects processors: 24-bit reverbs, COSM® modeling, RSS 3D effects plus standard JV’s Chorus and Reverb/Delay.
  • XV-3080 (2000) = XV-5080 without sample playback option, without COSM effects processor, smaller display.
  • XV-88 (2000) = keyboard version of XV-3080.
  • XV-5050 (2001) = XV-5080, without sample playback option, without SR-JV80 boards slots, polyphony reduced to 64, very small display. Size reduced to 1U, added USB support (editing via PC).
  • XV-2020 (2002) = XV-5050 in half rack module but no RSS effects, no COSM efx, no SR-JV80 boards slots, sound editing only via PC.

What was before JV-80?
JV-80 is based on PCM (Pulse Code Modulation) waveform playback. First of such made by Roland was model D-50 (1987), which became very popular. Not just only in the late 80’s, but also in 90’s (because of it’s analog synthesis emulation part which is quite powerfull – 4 oscillators per patch, nice smooth 12 dB resonant filter, 6 LFO’s, pulse width modulation). Next PCM synthesizer from Roland was U-110, which was later replaced by U-220 along with keyboard version labeled U-20. It was a very limited synthesizer with no filters of any kind, no assignable LFO’s, primitive pitch and vibrato adjustments (no envelope). The U-20 was in 1990 followed by U-50 which will be in the last minute renamed to D-70 due to popularity of D-50. D-70 had upgraded U-20 engine, some new waveforms and most importantly it added a resonant multimode filter. D-70 is definitely one of the most mysterious Roland synths, often overlooked and forgotten. The reason might be a bit hard user interface which has some impractical solutions that can make your life harder rather than easier. In parallel to D-70, Roland put out MV-30 which is very similar engine with added MC-50 sequencer. Finally in 1991 the JV-80 came out and this is where the legend began.

Quality issues with JV/XP series
At one point, in the mid 90’s, Roland switched to using SMD electrolytic capacitors. This has its benefits (gear has less weight) but drawbacks too (it can be harder to service). With that being said, it was discovered, first by users and then later confirmed by Roland themselves, that the electrolytic capacitors in Roland SR-JV80 expansion cards were not of good quality and by now (2019) many of them are failing. I have determined that the same capacitors were used at least in one XP synthesizer, model XP-50. Many of these caps have failed by now. In fact I have one of these myself and had to replace all of the SMD electrolytic capacitors. First symptoms were that audio would no longer work at the output. The good side of the story is, JV-1080 and JV-2080 owners are in a safe position as these actually use thru hole electrolytic capacitors. I can not confirm their quality level, but I never heard of any of these units failing due to bad capacitors. They are safe to use and operate for many years to come, which is something that can not be said for XP-50.

Some final words on the JV-80 vs JV-1080
They sound different due to 1) different digital filters 2) different anti alias filters.

  • Super JV has a filter from JD series (or a very close version of it). JV-880 has original filter from JV-80 series (also used in JV-90 and JV-1000). Emulation of that filter is possible with Super JV though it is less precise as you have less values to choose, particularly if you’re trying to emulate the “soft” resonance option from the JV. We discussed resonance compensation values above for both the hard and soft setting in the JV-80.
  • Antialiasing filter in Super JV is superior to the one in JV – which, depending on what kind of sounds you like is – welcome – or not so welcome feature. Mirroring in higher frequencies, particularly when using rich textures can fool the listener thinking the unit is 44kHz waveform set, though in reality it is not, it is 32kHz just like Super JV. I talk about mirroring above 16kHz which can happen during transposition, thought the waveforms are all 32 kHz.

Super JV was based on a far superior RISC processor which at that time was state of the art (sort of) hence the machine can take a lot of modulations real time, without sustaining damage on evelopes and LFOs – which again is welcome or not so welcome. This depends whether you prefer jumping envelopes as “more analog” while you tweak some parameter live on a synth. Which one should you buy? Well, JV-80 was really cool synth, however on your place i would go with 1080. I tested JV-1000 against Super JV and you can practically cover all of the JV sounds, minus aliasing artefacts! So for the harsh sound factor (alias abuse), or 100% authenticity, you will go JV-80/880 route, other than that look into 1080 or even better 2080 direction.

Roland experts
When it comes to experts in the Roland synthesizers that we covered in here, first name that comes in mind is of course Eric Persing. He used to post on a Gearslutz forum as a member “spectrum” and with a little help of the search tool one can find a real gold mine of valuable infos and resources. You can use this link to find some of his posts. Another name the comes to mind, especially about the nerdy details about ROM set and the Roland compression schemes it is definitely Edward from D-Tech. I highly suggest you visit his web page to learn more in-detail about the waveform ROM of these Roland romplers we have covered. Link here http://www.dtech.lv/techarticles_roland_exp.html