From Brian Willoughby Sent Sun, Feb 18th 2018, 08:03
Thanks. That makes a certain amount of sense. My only remaining question is how the TB-303 pulls off higher accuracy = with the same discrete R-2R DAC. Did Roland fine-tune the resistors on = each unit at the factory? Or are the TB-303 units simply not any more = accurate than the Avalon? Reading your message again, it seems that the DAC is only really = problematic at low frequency voltages. The DAC is fine enough at higher = frequency voltages but the VCO itself is where the problem lies. I know that the earlier Prophet 5 models - Rev 1 and Rev 2 - used = resistor ladders that were hand-calibrated to produce usable results. = The service manual warns that if a technician sees hand-soldered = resistors in parallel with the main resistors, they are NOT to remove = these. That=E2=80=99s how the accuracy was improved - by manually = tweaking the parallel resistances. I certainly wouldn=E2=80=99t want to = be involved in that sort of mess. The Rev 3 switched to a DAC chip with = many more bits so that the hand-tuning was no longer necessary. I=E2=80=99= m actually amazed at the massive differences between the Rev 2 and Rev 3 = Prophet 5 circuits. They all have software scaling to adjust for = individual VCO inaccuracies, based on using the CPU to measure the = actual frequency of each VCO for a given DAC CV output. Thanks again for the Avalon / TB-303 explanation, Brian. Brian On Feb 17, 2018, at 3:52 PM, Brian Castro <xxxxxxxxxxxx@xxxxx.xxx> = wrote: > Hi all, >=20 > I'll clear up any questions regarding the avalon cv range. >=20 > The avalon uses the same discrete R-2R DAC and 5.333V reference as the = TB-303. The DAC itself allows for a note range of MIDI Note 36 (C2) to = MIDI Note 99 (D#7). The 303/avalon analog circuitry is single supply and = the CV buffer is a single supply op amp that performs well down close to = the (-) rail. It starts to become non-linear below ~400mV, so the avalon = and tb-303 restrict the lowest note to MIDI Note 42 (F#2). Furthermore, = the 303/avalon VCO becomes increasingly out of scale at higher octaves = as the simple design does not incorporate any high-freq compensation. As = a result. the highest note is software-limited to MIDI Note 95 (B6). = This is equivalent to 4.917V.=20 > The note range could be extended in software but then lots of people = may complain about being out-of-tune. It's a give and take. > The external CV input is direct to the analog VCO and bypasses the = MCU, so it is not software-limited. The CV range is greater but will = become progressively out of tune. >=20 > On Thu, Feb 15, 2018 at 3:59 PM, Brian Willoughby = <xxxxxx@xxxxxxxxxxxx.xxx> wrote: >> Very interesting! >>=20 >> It=E2=80=99s an all-too-common design error to plug a 5 V DAC into a = system that can=E2=80=99t actually deliver 5 V because the supply rails = are running at exactly 5 V, too. There needs to be some headroom. >>=20 >> You might be able to fix this by replacing the op-amps in the Avalon = Bassline with a pin-compatible alternative that has better Rail-to-Rail = performance. This might be tricky for a couple of reasons. First, there = are many op-amp parameters that might change, so getting higher Voltages = may come at a sacrifice of some other performance. Second, this thing is = probably SMD, and you might not have heated tweezers. >>=20 >> I still haven=E2=80=99t looked at the TB303 schematic, but does = anyone know whether its supply runs at more than 5 V? >>=20 >> The 85 mV limit you=E2=80=99re seeing with the Avalon Bassline looks = suspiciously like an op-amp chip limit. But be warned that even the best = RR op-amps have a non-zero limit. >>=20 >> Brian >>=20 >> On Feb 15, 2018, at 3:37 PM, skkatter <xxxxxxxx@xxxxx.xxx> wrote: >> > I checked the Avalon Bassline CV out, 4.915 volts is the highest it = will go. The TT303's highest is 5 volts. >> > >> > And here's a short jam with it for good measure: = https://youtu.be/Hcf4eY5klf8 >> > >> > -Stephen >> > >> > On 14 February 2018 at 08:29, Brian Willoughby = <xxxxxx@xxxxxxxxxxxx.xxx> wrote: >> >> I guess that explains it, then. If it can=E2=80=99t produce as = high a CV as the TB303, then it can=E2=80=99t sequence the same high = notes. >> >> >> >> I=E2=80=99ll be curious to see whether you find that 5 V is the = limit. This sort of thing is the reason why I=E2=80=99ve passed over = many MIDI-to-CV products that were limited to 5 V. Sure, not every VCO = has more than 5 octaves of range, but those that do deserve to flex = their muscles. >> >> >> >> My new question is why the software emulations had the same limit. = Maybe there=E2=80=99s a completely separate reason and it=E2=80=99s just = a coincidence. >> >> >> >> Next time I have a minute, I hope to look over the TB303 = schematics to see how they handle CV generation from the sequencer. >> >> >> >> Brian >> >> >> >> >> >> On Feb 14, 2018, at 12:14 AM, skkatter <xxxxxxxx@xxxxx.xxx> wrote: >> >> > Heya Brian, >> >> > >> >> > Yes the Avalon does have CV and gate output from its internal >> >> > sequencer. I plugged it into one of the SEMs I happened to have = lying >> >> > around and its CV out also can't output the highest note. I'll = be able >> >> > to check the actual voltage coming out of it with a multimeter = this >> >> > evening after work, mine is currently hidden in a box somewhere = under >> >> > five hundred other boxes as I just moved house. :/ >> >> > >> >> > -Stephen >> >> > >> >> > On 14 February 2018 at 01:15, Brian Willoughby = <xxxxxx@xxxxxxxxxxxx.xxx> wrote: >> >> >> Does the Avalon have CV output from its internal sequencer? >> >> >> >> >> >> If so, measure the CV output to see if it matches the TB303 = output or if the Avalon CV generator is limited. You could either use a = Volt meter or just plug it into a known, good VCO (other than the = Avalon, which is clearly suspect). >> >> >> >> >> >> My untested hunch is that the Avalon VCO is totally analog and = can handle the high CV and reproduce the pitch, but the internal CV = generation can=E2=80=99t actually recreate the same voltages that the = TB303 is generating. Not sure why unless it=E2=80=99s going above 5 V, = which is a typical limit for a DAC without calibrated gain. A VCO = wouldn=E2=80=99t necessarily have a 5 V limit on its 1V/oct CV input, = especially not if it=E2=80=99s running on +/-10 V or higher rails. >> >> >> >> >> >> Brian >> >> >> >> >> >> On Feb 13, 2018, at 1:33 PM, skkatter <xxxxxxxx@xxxxx.xxx> = wrote: >> >> >>> TB303 and TT303 can play the highest C transposed up fine with = their >> >> >>> internal sequencers, Avalon cannot, however Avalon *can* play = the note >> >> >>> when you bypass it's internal sequencer and play it via = CV/Gate from >> >> >>> the TB303. Can anybody else re-create this behaviour? I = recorded a >> >> >>> little video to demonstrate: >> >> >>> >> >> >>> https://youtu.be/0Gbm-U1PZ7A >> >> >>> >> >> >>> -Stephen >> >> >>> >> >> >>> On 13 February 2018 at 21:31, DJ Maytag = <xxxxxxxx@xxxxxxxx.xxx> wrote: >> >> >>>> Even some software emulations had that limit, and it would = transpose an >> >> >>>> attempt at transposing E3 to E4 into C4 (its upper limit) = instead. >> >> >>>> >> >> >>>> On Tue, Feb 13, 2018 at 3:06 PM Computer Controlled = <xxxxxxx@xxxxx.xxx> wrote: >> >> >>>>> >> >> >>>>> Your past the limits of what it will play. I believe it=E2=80= =99s a 3 octave >> >> >>>>> limit. >> >> >>>>> >> >> >>>>> On Tue, Feb 13, 2018 at 10:12 AM, skkatter = <xxxxxxxx@xxxxx.xxx> wrote: >> >> >>>>>> Avalon users! Sorry for the spam, have a quick technical = question. >> >> >>>>>> >> >> >>>>>> On my Avalon if I program a pattern with the upper C, and = have that >> >> >>>>>> upper C up one octave (or lower C up two octaves) it plays = fine. >> >> >>>>>> However, if I then transpose that pattern up, the upper C = cuts out, >> >> >>>>>> all the other notes will play fine however. >> >> >>>>>> >> >> >>>>>> Is this normal Avalon Bassline behaviour? Because it isn't = normal >> >> >>>>>> TB303 behaviour. >> >> >>>>>> >> >> >>>>>> Thanks, >> >> >>>>>> Stephen 303 problems today.