From 5c377b081415aff0040dc56116a9a4226c2ac295 Mon Sep 17 00:00:00 2001 From: Chris Xiong Date: Fri, 1 Jul 2022 03:28:28 +0800 Subject: Fix table captions, add a few more photos. --- blog/post/2019-04-25.html | 2 +- blog/post/2020-11-20.html | 23 +++++++++++++++++++---- 2 files changed, 20 insertions(+), 5 deletions(-) (limited to 'blog/post') diff --git a/blog/post/2019-04-25.html b/blog/post/2019-04-25.html index ca48f89..af8b757 100644 --- a/blog/post/2019-04-25.html +++ b/blog/post/2019-04-25.html @@ -77,7 +77,7 @@ for(let x of cl) a.setAttribute('x',n-1); a.innerHTML='show'; a.href='javascript:void(0)'; - b.innerHTML=dd.getAttribute('caption'); + b.innerHTML=dd.getAttribute('data-caption'); s.appendChild(b); s.appendChild(document.createTextNode(' [')); s.appendChild(a); diff --git a/blog/post/2020-11-20.html b/blog/post/2020-11-20.html index 245b88d..bd6f0c9 100644 --- a/blog/post/2020-11-20.html +++ b/blog/post/2020-11-20.html @@ -77,7 +77,7 @@ for(let x of cl) a.setAttribute('x',n-1); a.innerHTML='show'; a.href='javascript:void(0)'; - b.innerHTML=dd.getAttribute('caption'); + b.innerHTML=dd.getAttribute('data-caption'); s.appendChild(b); s.appendChild(document.createTextNode(' [')); s.appendChild(a); @@ -323,8 +323,22 @@ for(let x of al)

And then something extremely stupid happened. Any proper electrical engineer will cringe hard. At this point I was getting cocky, and started randomly probing around with my multimeter on the volume board. I “accidentally” shorted the first two pins of the connector going from the main board into the volume board (pin 1 and pin 2 on CN7 of volume board). The output from the headphone jack immediately turns into complete garbage (severe distortions on low frequencies). The OpAmps chip on the volume board started getting ridiculously hot… crap! I still managed to fry something for an otherwise “perfect” modding project!

Of course this is undesirable. So I had to find a fix to this.

Fortunately, nothing on the main board seemed to be hurt. I can just bypass the volume control and get the correct output on output 1 using some jumper wires. So the fault is contained in the volume board. I’ve basically sent -15V straight into the base of two transistors, but measuring those transistors didn’t reveal anything wrong with them. So I had to assume I have fried the amp chip (NJM4565). I got a few replacements (NJM4580, compatible spec-wise) from taobao, and replaced the “faulty” NJM4565 (still without using flux). But the audio output is still messed up and the opamp chip still gets very hot after the replacement. I decided to give up for now, and look into the thing later. Meanwhile I just tucked some of the wires from CN6 into CN7 so that I can still get analog output from output 1 on the back panel.

+ + + + + +
+
+
My terrible SMD soldering +
+		 +
+
Recombined plug for volume board bypass +
+
-
My terrible SMD soldering +

Recording Setup Update Part 1

Since I have always wanted to record my SD-80 through a digital link (even before frying the analog output), I have been keeping an eye on cheap digital recording solutions. Modern professional audio interfaces never come with digital input on budget models. Among the older interfaces, UA-25(EX) from Roland (EDIROL / Cakewalk) seems to be a reasonable choice. There are also a bunch of different models from various brands of the same era that have digital inputs. However these models are virtually impossible to get in China. Then some cheapo consumer grade stuff caught my attention – several relatively nameless brands have “sound cards” for home theater uses that have digital I/O. Those are priced at roughly 200~300 Chinese yuan. Among those I found a more widely recognized brand called Terratec. They have PCI-e and USB sound cards that comes with digital I/O and are available for purchase in China. Price are on the higher end (300+ CNY), however still way cheaper than the cheapest professional audio interface that doesn’t have digital I/O (those start from ~800 CNY). Plus these models seem to have a reasonable Linux user base, so I got their Aureon 7.1 USB.

@@ -1360,7 +1374,7 @@ Reverb, Chorus, MFX (16 slots, 67 types), EQ, compressor (drum part), Surround,
  • Acoustic drum set from the solo set is a cut down version of the studio kit from SRX-03 Studio, which is also the source of Super Quartet’s drums.
  • Piano patches are pulled straight from SR-JV80-09, which is also included in SRX-07. SC-8850 has the same Piano sound.
  • Clavi is almost identical to one of the many clavi patches from SC-8820/8850, and is likely ultimately from the JVs and SR-JV80 boards.
    • -
  • Samples of Flute vib sounds identical to the one with the same name (“Flute Vib3 A”) in SRX-03.
    • +
  • Samples of Flute vib sound identical to those with the same name (“Flute Vib3 A/B/C”) in SRX-03.
  • Samples of St.Brass and St.Sm Choir also come straight from SRX-03.
  • Multiple sound effects are from earlier SC models. Some are also used by XV-5080’s GM2 mode.
  • A lot of patches in the special sets are pulled from the XV-5080. They use the exact same parameters, except the waveforms. If you can find a preset with the same name as an instrument from SD-80’s special set in the XV-5080, chances are they sound almost identical, especially since a lot of them are analog/digital synth patches, and waveforms don’t matter as much. There are a few exceptions – a preset with the name “Cascade” is found in both instruments, but they have nothing in common except the name. There are also a lot of XV-5080 “inspired” patches: they have different names from the original XV-5080 patch, but very similar sound design. In fact, the “Cascade” patch mentioned above is one of these XV-5080 “inspired” patch, but you have to figure out the original yourself as I forgot which one it is.
    • @@ -1512,6 +1526,7 @@ Stereo Auto Wah
  • Very few (if any) preset patches uses the modulation matrix of the XV engine correctly. All of them has the modulation source set to ‘OFF’.
  • Only 5 of all preset patches used non-default tone structures: “Runaway Rez”, “Purple Spin”, “FM layer”, “FM Delight”, and “Xmod EP”. All of them are in the special sets. 3 of them are unmodified XV-5080 patches.
  • There doesn’t seem to be a way to set the system tempo of the SD-80/90 with MIDI messages, nor can the SD-80/90 sync its MIDI clock with a host, rendering the system clock mostly useless. Neither of these two is true for the XV-5080.
    • +
  • Ever wondered why some patches have seemingly nonsensical waveforms selected in disabled tones [23]? Just look up those wave numbers in the waveform list of XV-5080 or the corresponding SRX board! [24] This, once again, suggests Roland used the XV-5080 as the development platform for the StudioCanvas.
  • From Sound On Sound’s review of the SD-90: “To me, however, USB audio and the Sound Canvas sound set don’t add up to £799, and although I grew to like the SD90, I’m not sure how many people will find it attractive at this price point.” – ZUN, apparently.

    • Errata of the original post

    @@ -1533,7 +1548,7 @@ Stereo Auto Wah
    -
    [1]: Judging by the way Roland utilized the SH-3 CPU in MC-909, which has a 16MHz external clock input and a 8x multiplier (128MHz internal clock), I would guess the CPU in SD-80 also works at 8x multiplier and therefore 96MHz internally.
    [2]: Later the source of this DC bias is determined to be SD-80 itself, not the recording device. See the next section.
    [3]: オールインワン・モデルSD-90でご好評いただいた、新開発MIDI音源部を搭載したマルチティンバー音源が登場。 As seen here. I don’t actually know any Japanese and just pieced stuff together randomly. Sorry if I butchered your language.
    [4]: Munt isn’t strictly an emulation. It doesn’t emulate the CPU or actual circuitry of the MT-32. See below.
    [5]: without Roland losing their mind and releasing all internal documentation on the XV engine, or some absolute madlad spending 15 hours everyday on reverse engineering the thing for half a year, that is.
    [6]: SD-80 has 32MiB of compressed wave ROM, see the “list of integrated circuit chips on SD-80 main board” in the first section. Roland’s waveform compression scheme usually results in a ~50% compression ratio. Therefore the content is roughly equal to 64 MiB of uncompressed 16-bit PCM wave.
    [7]: The “23C128” kind of gave it away – they are the μPD23C128040ALGY mask ROM chips from NEC, which is the exact same type of ROM used in XV-5080. Unlike the XV-5080 though, the SD-80 makes use of both its J variant and K variant, while the XV-5080 only uses the J variant (these variants have symmetric pin configuration).
    [8]: Well, the Raspberry Pi isn’t really suitable for this task because it doesn’t have enough GPIO pins. But there’s an easy workaround for that.
    [9]: For readers who wonders what “mask” means in this context: you can treat a mask ROM as a huge array of tiny switches that can’t be turned on or off once manufactured. You can access the state of a group of switches by giving an address to its input pins. The mask is used as a template of the states of these switches during the manufacture process. This is electrical engineering amateur Chris trying to explain mask ROM in layman’s terms.
    [10]: HyperCanvas (HQ-GM2) or Cakewalk TTS-1, which is a rebranding of the former; SuperQuartet (HQ-QT) and Orchestral (HQ-OR). A plugin called GrooveSynth (P5antom) bundled with several earlier Cakewalk products providing patches from the MC-303 Groovebox also uses this engine.
    [11]: which is kind of weird considering Spectrasonics basically spun off from Roland
    [12]: Also used in XP-80, see the errata section of its service manual.
    [13]: HD64F7017F28, SH7017 in parts list
    [14]: 40 of the 90 types will take up all three slots, most likely due to the reduced DSP power.
    [15]: Support for multisamples also exist in Fantom S/S88, so this is more likely due to an updated system software rather than changes of the synth engine.
    [16]: There are evidence that some of them are handled by software (SD-80 having one more LFO per part than the XV-5080). However it can also be using LFO blocks in the XV chip that is unused in the XV-5080.
    [17]: This mapping is used for DT1/RQ1 system exclusive messages.
    [18]: The first half is also true for earlier SC models (SC-55 <-> JV-880, SC-88 <-> JV-1080, SC-88Pro <-> JV-2080, SC-8850 <-> XV-3080). However the second half isn’t. Earlier SC models employs a GS-specific address map which looks nothing like their counterparts.
    [19]: And also the last time, since neither the SD-20 nor the SD-50 has such editability.
    [20]: Apparently it’s from the SC-8850.
    [21]: The multisample from Roland Cloud seem to have an extra sample in the highest register, which sounds like it’s processed with a low-pass filter with very low cut off frequency and makes it sound like garbage. This is also the case for the version included in the original SRX-09 boards.
    [22]: The owner’s manual of the SD-80 contains blatant lies. It says “It is not possible for the edited sounds to be saved in the internal memory of the SD-80” (which is directly copied from SD-90’s manual), and goes on to teach you how to save a user patch.
    +
    [1]: Judging by the way Roland utilized the SH-3 CPU in MC-909, which has a 16MHz external clock input and a 8x multiplier (128MHz internal clock), I would guess the CPU in SD-80 also works at 8x multiplier and therefore 96MHz internally.
    [2]: Later the source of this DC bias is determined to be SD-80 itself, not the recording device. See the next section.
    [3]: オールインワン・モデルSD-90でご好評いただいた、新開発MIDI音源部を搭載したマルチティンバー音源が登場。 As seen here. I don’t actually know any Japanese and just pieced stuff together randomly. Sorry if I butchered your language.
    [4]: Munt isn’t strictly an emulation. It doesn’t emulate the CPU or actual circuitry of the MT-32. See below.
    [5]: without Roland losing their mind and releasing all internal documentation on the XV engine, or some absolute madlad spending 15 hours everyday on reverse engineering the thing for half a year, that is.
    [6]: SD-80 has 32MiB of compressed wave ROM, see the “list of integrated circuit chips on SD-80 main board” in the first section. Roland’s waveform compression scheme usually results in a ~50% compression ratio. Therefore the content is roughly equal to 64 MiB of uncompressed 16-bit PCM wave.
    [7]: The “23C128” kind of gave it away – they are the μPD23C128040ALGY mask ROM chips from NEC, which is the exact same type of ROM used in XV-5080. Unlike the XV-5080 though, the SD-80 makes use of both its J variant and K variant, while the XV-5080 only uses the J variant (these variants have symmetric pin configuration).
    [8]: Well, the Raspberry Pi isn’t really suitable for this task because it doesn’t have enough GPIO pins. But there’s an easy workaround for that.
    [9]: For readers who wonders what “mask” means in this context: you can treat a mask ROM as a huge array of tiny switches that can’t be turned on or off once manufactured. You can access the state of a group of switches by giving an address to its input pins. The mask is used as a template of the states of these switches during the manufacture process. This is electrical engineering amateur Chris trying to explain mask ROM in layman’s terms.
    [10]: HyperCanvas (HQ-GM2) or Cakewalk TTS-1, which is a rebranding of the former; SuperQuartet (HQ-QT) and Orchestral (HQ-OR). A plugin called GrooveSynth (P5antom) bundled with several earlier Cakewalk products providing patches from the MC-303 Groovebox also uses this engine.
    [11]: which is kind of weird considering Spectrasonics basically spun off from Roland
    [12]: Also used in XP-80, see the errata section of its service manual.
    [13]: HD64F7017F28, SH7017 in parts list
    [14]: 40 of the 90 types will take up all three slots, most likely due to the reduced DSP power.
    [15]: Support for multisamples also exist in Fantom S/S88, so this is more likely due to an updated system software rather than changes of the synth engine.
    [16]: There are evidence that some of them are handled by software (SD-80 having one more LFO per part than the XV-5080). However it can also be using LFO blocks in the XV chip that is unused in the XV-5080.
    [17]: This mapping is used for DT1/RQ1 system exclusive messages.
    [18]: The first half is also true for earlier SC models (SC-55 <-> JV-880, SC-88 <-> JV-1080, SC-88Pro <-> JV-2080, SC-8850 <-> XV-3080). However the second half isn’t. Earlier SC models employs a GS-specific address map which looks nothing like their counterparts.
    [19]: And also the last time, since neither the SD-20 nor the SD-50 has such editability.
    [20]: Apparently it’s from the SC-8850.
    [21]: The multisample from Roland Cloud seem to have an extra sample in the highest register, which sounds like it’s processed with a low-pass filter with very low cut off frequency and makes it sound like garbage. This is also the case for the version included in the original SRX-09 boards.
    [22]: The owner’s manual of the SD-80 contains blatant lies. It says “It is not possible for the edited sounds to be saved in the internal memory of the SD-80” (which is directly copied from SD-90’s manual), and goes on to teach you how to save a user patch.
    [23]: For example, nearly all acoustic bass patches have a disabled tone with wave number 249 “TenBlwSaxVib” selected, and the Fiddle 2 vib patch have a disabled tone with wave number 276 “Blow Pipe” selected.
    [24]: Wave #249 in XV-5080 is UprightBs 2A, and Wave #276 in SRX-09 is Fdl Pizz 1C (Fiddle Pizzicato).