Updated SSI2130 VCO Datasheet available...

[neil.johnson]

Power rails of the SSI2130 - just to be sure.

Neil

[neil.johnson]

Also note that for functionality, I would recommend 2V above V- to 1.5V below V+ should be fine.

Also worth pointing out is that the PWM comparator is actually a rather good comparator with very good performance (way better than is needed for PWM). It could be used for other things besides PWM.

Neil

[bence]

thank you again!:)

i plan to clamp with 10k and a 3v zener, so that should keep the input in that range.
you also mentioned adding some hysteresis - the datasheet suggests that the comparator already has hysteresis, but it is not specified.
anyway on the board i'm drawing i've put 10k on the +ve input and 1M feedback as a starting point, do you see any problem with this?

other things besides pwm:
so for a tz vco without pwm it could replace the external comparator for even lower parts count?
(and maybe 2.5V is enough for a high level on the time reverse pin?)
edit: ok, the mosfet needs a bit more gate voltage than 2.5V, but maybe it could be replaced with a bjt?

[neil.johnson]

thank you again!:)

i plan to clamp with 10k and a 3v zener, so that should keep the input in that range.
you also mentioned adding some hysteresis - the datasheet suggests that the comparator already has hysteresis, but it is not specified.
anyway on the board i'm drawing i've put 10k on the +ve input and 1M feedback as a starting point, do you see any problem with this?

other things besides pwm:
so for a tz vco without pwm it could replace the external comparator for even lower parts count?
(and maybe 2.5V is enough for a high level on the time reverse pin?)
edit: ok, the mosfet needs a bit more gate voltage than 2.5V, but maybe it could be replaced with a bjt?

There is no hysteresis on the internal comparator - you need to add that yourself. For basic PWM functionality you can get away without it, but a little hysteresis can improve the signal quality.
I don't think you can use a BJT as you really need the very low Rds(on) of the MOSFET to correctly operate the opamp circuit. You might be able to drive the MOSFET from a BJT, but then you're soon in the same board area and cost of using an external comparator.

Cheers
Neil

[neil.johnson]

Important update: unused mixer control inputs must be left unconnected. DO NOT CONNECT TO GROUND!!!

Otherwise...

Neil

[neil.johnson]

Oooh, one other interesting result for the tuning stability aficionados: I ran a test to see how good the tuning is over a wide range. With CV of 0V to 11V (in 1V steps) I found the SSI2130 can get to within less than +/- 1 cent out. That's from around 8 Hz up to around 17 kHz. Over a narrower range of 8 octaves it is less than 0.6 cent out.

Sweet

Neil

[Stab Frenzy]

I've run into a peculiar quirk when testing my prototype on a variety of Eurorack power supplies. I have one supply where the -12V rail comes up around 8ms after the +12V, using this supply the 2130 locks up with the voltage on tcap at around 3.5V. I'm not using the Strobe pin in this design so it's not broken out, otherwise I'd see if I could reset it after both rails were up. The 2130 is supplied by +5V and -5V regulators, the -5V rail floats at around +0.7V until the -12V rail is up.

Any thoughts on what's going on internally and how it might be avoided? This situation isn't that unlikely in the wild west of Eurorack power so I'd like to design the module in a way that can tolerate it on startup.

[KSS]

@neil.johnson
^Your description is log conformance rather than tuning stability. Conformance means it can hit those numbers. Stability requies it hold them through time and temperature changes. Both important, but not the same thing.

[thx2112]

Important update: unused mixer control inputs must be left unconnected. DO NOT CONNECT TO GROUND!!!

Otherwise...

Neil

Came here to say this. Was wondering where the short on -5V was...

[neil.johnson]

Important update: unused mixer control inputs must be left unconnected. DO NOT CONNECT TO GROUND!!!

Otherwise...

Neil

Came here to say this. Was wondering where the short on -5V was...

Yeah, the datasheet is going to be updated RSN

Neil

[DozenCrows]

Is there any risk of harm when using a higher current than recommended on the mixer control pins (within a 'sensible' range, i.e. not hundreds of mA)? I got a control scheme working based on 100uA, before the datasheet was revised to recommend 50uA. My scheme works, and I don't see any distortion on the output.

On a couple of the mixer control pins I have a series diode before the resistor, so those pins are effectively floating when set to "no output"; will there be any issue with that?

[neil.johnson]

Is there any risk of harm when using a higher current than recommended on the mixer control pins (within a 'sensible' range, i.e. not hundreds of mA)? I got a control scheme working based on 100uA, before the datasheet was revised to recommend 50uA. My scheme works, and I don't see any distortion on the output.

We found that the higher currents into the mixer pins could cause minor pitch errors. At 50uA per mixer control pin the worst case error is minimal.

On a couple of the mixer control pins I have a series diode before the resistor, so those pins are effectively floating when set to "no output"; will there be any issue with that?

That's fine - the default is float when not used. Note that real diodes suffer reverse leakage so there will be some current flow, just very very tiny.

Neil

[DozenCrows]

Thank you Neil.

We found that the higher currents into the mixer pins could cause minor pitch errors. At 50uA per mixer control pin the worst case error is minimal.

I'll revise my scheme in that light, to be on the safe side! Looks like only changing a few resistor values is what's needed.

The datasheet recommend using 50kOhm as the minimum resistance on these pins, but the example cross-fade circuit uses 20kOhm. Is there some trade-off with these values as well, given that the recommended 50uA can be achieved with a pretty wide range of resistor and voltage values?

That's fine - the default is float when not used. Note that real diodes suffer reverse leakage so there will be some current flow, just very very tiny.

Understood. As I'm not actually going to silence, any such leakage shouldn't have an audible effect.

[neil.johnson]

The crossfade circuit puts out 0 to 1V control voltages so 20k gives 50uA. If you adapt it for 5V control then you need to change the mixer resistors to 100k.

Neil

[de calvus]

Greetings, my first Muffs post - have been lurking for some time, but now am getting out of the woodwork as I’m trying to get a better understanding of the thru-zero circuit from the datasheet.

When simulating it with a +-5V (5Vp) modulating sine signal the current into the LIN FREQ pin (I_Lin on the graph below) is ‘asymmetrical’. What I mean is it seems the amplitude of I_Lin corresponding to the negative part of the sine input is smaller than for the positive part, and also the current goes below the 5uA reference-current and down to about -1uA before it pops up again)

It’s my first experience with thru-zero VCO’s so am a bit unsure how it’s supposed to work. Can someone confirm this is the correct operation?

I had thought the curve would be symmetrical, like a full-wave rectified signal, with the bottom of the curve starting at 5uA.

The datasheet (page 5) mentions that a reference-current of about 5uA should be supplied to the LIN FREQ pin (from Vref=2.5V through a 500k resistor). This is handled by level-shifting the output of the first op amp (IC2a) to -2.5V (when the modulating input signal is 0V). Since the mosfet is turned on IC2b operates in inverting configuration and the -2.5V on it’s input then gives a +2.5V output, which R1=500k converts to the needed 5uA current
Because the comparator gets -2.5V on the inverting input when the modulating input signal is at it’s 0V center, the comparator output (Time Reverse signal) is not a 50% duty cycle square (as it would be in a zero-crossing detector), which is the reason for the ‘asymmetrical’ I_Lin current.
-

SSI2130_sim10.jpg (39.5 KiB) Viewed 254 times

If the current is supposed to be symmetrical (which is the part I’m not sure about) wouldn’t it be better to use a 500k resistor directly on the LIN FREQ pin (R11 on the below schematic) for supplying the 5uA. And then leave out the level-shift resistor R4 on above schematic. This would give a symmetrical current since the comparator correctly detects the zero-crossing of the input
The gain of the circuit is determined by R2/R3, but also by R1, which for example could be 500k or a lower value
-

SSI2130_sim11.jpg (38.46 KiB) Viewed 254 times

the small glitches on the I_Lin graph seem to be caused by the 2N7000 mosfet, but is probably because of a not optimal spice simulation model of the Fet. When replaced by an ideal switch the simulation showed no glitches, so I guess they are not present in the real circuit

[KSS]

Welcome to MuffWiggler, de calvus!

[de calvus]

Thanks, nice to be here

[neil.johnson]

Greetings, my first Muffs post - have been lurking for some time, but now am getting out of the woodwork as I’m trying to get a better understanding of the thru-zero circuit from the datasheet.

When simulating it with a +-5V (5Vp) modulating sine signal the current into the LIN FREQ pin (I_Lin on the graph below) is ‘asymmetrical’. What I mean is it seems the amplitude of I_Lin corresponding to the negative part of the sine input is smaller than for the positive part, and also the current goes below the 5uA reference-current and down to about -1uA before it pops up again)

It’s my first experience with thru-zero VCO’s so am a bit unsure how it’s supposed to work. Can someone confirm this is the correct operation?

Yes that is correct operation.

I had thought the curve would be symmetrical, like a full-wave rectified signal, with the bottom of the curve starting at 5uA.

Nope.

The datasheet (page 5) mentions that a reference-current of about 5uA should be supplied to the LIN FREQ pin (from Vref=2.5V through a 500k resistor). This is handled by level-shifting the output of the first op amp (IC2a) to -2.5V (when the modulating input signal is 0V). Since the mosfet is turned on IC2b operates in inverting configuration and the -2.5V on it’s input then gives a +2.5V output, which R1=500k converts to the needed 5uA current
Because the comparator gets -2.5V on the inverting input when the modulating input signal is at it’s 0V center, the comparator output (Time Reverse signal) is not a 50% duty cycle square (as it would be in a zero-crossing detector), which is the reason for the ‘asymmetrical’ I_Lin current.

That is all normal.

If the current is supposed to be symmetrical (which is the part I’m not sure about) wouldn’t it be better to use a 500k resistor directly on the LIN FREQ pin (R11 on the below schematic) for supplying the 5uA. And then leave out the level-shift resistor R4 on above schematic. This would give a symmetrical current since the comparator correctly detects the zero-crossing of the input

If you do that then in effect you set the carrier frequency to 0.

When replaced by an ideal switch the simulation showed no glitches, so I guess they are not present in the real circuit

Beware of simulations, they good to a point, but often that point is not what you want.

Neil

[de calvus]

Thanks Neil, that’s good to know. I will happily continue with the PCB layout then:-)

[marcodellav]

I'm wondering if something like this: https://plastronics.com/catalog/standard-qfn-clamshell/ is available for purchase anywhere. Going to search around. Might be nice for prototyping or DIY with these QFN ICs.

all manner of qfn soldering videos on youtube, saw one with someone using a home etched pcb with no solder mask, looks like there's one technique where you hold the iron against the underside of the pcb to melt the solder on the central pad (pcb was designed with no copper or mask under that part by the looks)

A friend of mine gave me some tips on this, but also said it may not be worth the effort. I mean, might as well just order 50-100 chicklet adapters from a pick and place shop with the IC already on them.

Hey not sure if you're still searching for a solution for these chips but I bought a couple of these QFN-32 to DIP-36 SMT Adapters (0.4 mm pitch, 4.0 x 4.0 mm body) from proto advantage. They should be in soon, and they come in a kit with stencil and solder paste. I've been looking for some cheapo Chinese sockets but I couldn't find the 0.4 mm pitch versions, only 0.5 so I decided to bite the bullet and try with these. I can let you know how it goes if you're interested.

https://www.proto-advantage.com/store/p ... id=3000132

[J3RK]

I'm wondering if something like this: https://plastronics.com/catalog/standard-qfn-clamshell/ is available for purchase anywhere. Going to search around. Might be nice for prototyping or DIY with these QFN ICs.

all manner of qfn soldering videos on youtube, saw one with someone using a home etched pcb with no solder mask, looks like there's one technique where you hold the iron against the underside of the pcb to melt the solder on the central pad (pcb was designed with no copper or mask under that part by the looks)

A friend of mine gave me some tips on this, but also said it may not be worth the effort. I mean, might as well just order 50-100 chicklet adapters from a pick and place shop with the IC already on them.

Hey not sure if you're still searching for a solution for these chips but I bought a couple of these QFN-32 to DIP-36 SMT Adapters (0.4 mm pitch, 4.0 x 4.0 mm body) from proto advantage. They should be in soon, and they come in a kit with stencil and solder paste. I've been looking for some cheapo Chinese sockets but I couldn't find the 0.4 mm pitch versions, only 0.5 so I decided to bite the bullet and try with these. I can let you know how it goes if you're interested.

https://www.proto-advantage.com/store/p ... id=3000132

That would be cool, please do let us know how it goes! I still haven’t gotten fully back up to speed yet, so I haven’t made any further progress with this yet, but will be eventually.

[anotherlin]

Have you guys checked Sound Semiconductor's website lately?
It seems there's a new SSI2131 VCO available. Less features than the already available SSI2130, but probably more "useful" as it has only "what's needed". And in a less scary (to solder) SOP-16 package!

Regarding the SMD to DIP adapters, I've done my homework: https://github.com/anotherlin/smd2dip

Seems https://www.cesyg.co.uk/ is only shipping to UK. Would it be possible for just one exception to France?
Neil Johnson, I'm sending you a message.

[neil.johnson]

Have you guys checked Sound Semiconductor's website lately?
It seems there's a new SSI2131 VCO available. Less features than the already available SSI2130, but probably more "useful" as it has only "what's needed". And in a less scary (to solder) SOP-16 package!

Yes it is quite a sweet little brother to the full-featured SSI2130. Easier to handle package, yet retains many of the key features (wide expo range and stability, PWM, hard/soft sync, etc) that make the SSI2130 a great VCO.

Not yet stocked by distributors but coming soon.

Seems https://www.cesyg.co.uk/ is only shipping to UK. Would it be possible for just one exception to France?
Neil Johnson, I'm sending you a message.

Trying not to say anything political, but let's just say that the paperwork and costs with shipping out of the UK to Europe is not something I feel like coping with at the moment.

Cheers
Neil

[anotherlin]

Trying not to say anything political, but let's just say that the paperwork and costs with shipping out of the UK to Europe is not something I feel like coping with at the moment.

No problem, I've just ordered from the other authorized UK dealer. Hope everything will go well.
The paperwork must be a real pain in the $%$#% that you don't want to sell to me!

[adam]

so, how about that new DAB2130 breakout module then?