A means of closing a JFET switch with a postive voltage?

[CDRIVE]

OK, I feel like I'm shoveling sh!t against the tide. The 9V battery limitation of this project leaves zero overhead to play with. It presented issues in your first topic, which is really the starting line of this topic. This has been a royal PITA right out of the gate... Pun? As I stated earlier I've spiced quite a few S&H schemes but your 9V battery has been a quagmire from the get go. I'm burning out!

Chris

 

[(*steve*)]

yeah that sounds odd to me too. Most modular synth stuff I've seen runs from +/-15V.

is the reason for this 9V battery because you felt you needed a completely independent supply for the fet?

 

[AFex54]

yeah that sounds odd to me too. Most modular synth stuff I've seen runs from +/-15V.

is the reason for this 9V battery because you felt you needed a completely independent supply for the fet?

its not a modular synth, its a MFOS Noise Toaster, I didnt design it.
Im just adding trying to add a keyboard so it can be pre-tuned.

 

[CDRIVE]

yeah that sounds odd to me too. Most modular synth stuff I've seen runs from +/-15V.

is the reason for this 9V battery because you felt you needed a completely independent supply for the fet?

its not a modular synth, its a MFOS Noise Toaster, I didnt design it.
Im just adding trying to add a keyboard so it can be pre-tuned.

That much I've got nailed down but the 9V source limitation with no overhead to play with is proving to be an exercise in futility. That plus lack of a negative supply rail, that many S&H circuits employ, makes this circuit a hard nut to crack.

Chris

 

[(*steve*)]

I'll have to read the whole thread later (tomorrow) and get myself fully up to speed.

I'll be looking for stuff including why there's only a single ended 9v supply.

 

[AFex54]

I'll have to read the whole thread later (tomorrow) and get myself fully up to speed.

I'll be looking for stuff including why there's only a single ended 9v supply.

it was designed to be a low cost, Lo-Fi noise maker that a beginner would build themselves.
it is by no means a proper synth, and it can run off 9V battery for hours, like 10mA consumption.
you probably wont find many answers in this thread ,as CDRIVE pointed out earlier Ive unneccesarily started multiple threads on the same subject of this project.

 

[AFex54]

and ive managed to buy the wrong op amp again,
the input bias current of this LM6144 is charging the sampling cap like it did with LM324 except 10 times worse .
I dont even understand the datasheets so whatever i think I may know about a given component is probably wrong.
http://docs-europe.electrocomponents.com/webdocs/077f/0900766b8077ffc2.pdf
I think this AD822ANZ could work but of course it mentions input bias current about 10 times with different values so i wouldnt count on it.
the input bias current seems to be in picoamps,
I know the TL072 has input bias current of 200 pA and can sample and hold effectively

 

[(*steve*)]

OK, so I've been thinking about this while I've been sleeping (and I haven't yet read back through your thread(s)). Let me know if I'm right.

• You have a keyboard with some undefined number of keys.
• Each key will correspond to an individually adjusted voltage.
• On the press of a key, the voltage associated with that key appears at the output.
• On the release of the key, the voltage remains constant at the output.
• The circuit should run from a single ended 9V supply (a 9V battery).
• The voltage output should vary from 0V to as close as possible to 9V.

Additionally, can we assume the following?

• The startup conditions are undefined until the first key is pressed.
• Pressing 2 keys simultaneously produces an undefined output.

"undefined" simply means any valid output voltage or a high impedance output state.

If these are reasonable then I have a solution for you.

 

[CDRIVE]

Steve, keep a lot of coldies on hand. It's a long read if you go all the way back. I put links in this topic that will take you to beginning of this saga. Come to think of it you may want some tuck before you get through it all.

Chris

 

[Alec_t]

Why do you even want a sample-and-hold circuit? Surely you simply need to read the voltage defined by the key states periodically (say every few mS) and play the appropriate sound until that voltage changes?

 

[AFex54]

OK, so I've been thinking about this while I've been sleeping (and I haven't yet read back through your thread(s)). Let me know if I'm right.

• You have a keyboard with some undefined number of keys.
• Each key will correspond to an individually adjusted voltage.
• On the press of a key, the voltage associated with that key appears at the output.
• On the release of the key, the voltage remains constant at the output.
• The circuit should run from a single ended 9V supply (a 9V battery).
• The voltage output should vary from 0V to as close as possible to 9V.

Additionally, can we assume the following?

• The startup conditions are undefined until the first key is pressed.
• Pressing 2 keys simultaneously produces an undefined output.

"undefined" simply means any valid output voltage or a high impedance output state.

If these are reasonable then I have a solution for you.

yes, thats it.
its monophonic.
BTW the VCO either doesnt oscillate or produces infrasonic frequencies below about 1V so getting absolute zero isnt vital

 

[AFex54]

Why do you even want a sample-and-hold circuit? Surely you simply need to read the voltage defined by the key states periodically (say every few mS) and play the appropriate sound until that voltage changes?

Im not sure what you mean by that but I havent consider any other method than a sample and hold circuit, I originally seen the sample and hold in the MFOS CV keyboard schematic so i assumed that was the only way,
my keyboard is a lot different to MFOS keyboard so its very possible there is an easy/different way of doing this

 

[(*steve*)]

yes, thats it.

OK, I'll draw you a circuit -- I'll only draw it for 2 keys, but you'll be able to see how to extend it for as many as you like (I hope)

 

[(*steve*)]

OK, here we go...



The parts in red are required for each key.

Yeah, I know it looks a lot, but it's not really.

In the top circuit:
- U8 is a Schmitt trigger inverter. These come 6 to a package, so you only need 1 of these IC's (40106) for each 6 keys.
- U2 is a D flip flop, or any of a number of latches. You can get 8 of these to a package, so 1 IC per 8 keys (40373, 40374).
- S2 is a CMOS switch, there are 4 of these in a package, so only 1 IC per 4 keys.

The circuit at the top creates a brief pulse that is delayed slightly. This ensures that everything is set up to latch the switch position (and fixes problems with switch bounce.

The advantage of this circuit is that the keys will not be a source of noise in the circuit and the voltage will remain fixed indefinitely. It will not drift as a sample and hold will.

Note that the output is relatively high impedance. If you have a rail to rail op-amp, you could use this to buffer the output voltage.

The C and R in the key circuit can be 0.1uF and 1MΩ.
The Cs and Rs in the delay circuit can be 0.1uF and 100kΩ.
R6 could be 100kΩ
The trimpots could be 10kΩ
The diodes can be 1N4148

 

[AFex54]

OK, here we go...

View attachment 21531

The parts in red are required for each key.

Yeah, I know it looks a lot, but it's not really.

In the top circuit:
- U8 is a Schmitt trigger inverter. These come 6 to a package, so you only need 1 of these IC's (40106) for each 6 keys.
- U2 is a D flip flop, or any of a number of latches. You can get 8 of these to a package, so 1 IC per 8 keys (40373, 40374).
- S2 is a CMOS switch, there are 4 of these in a package, so only 1 IC per 4 keys.

The circuit at the top creates a brief pulse that is delayed slightly. This ensures that everything is set up to latch the switch position (and fixes problems with switch bounce.

The advantage of this circuit is that the keys will not be a source of noise in the circuit and the voltage will remain fixed indefinitely. It will not drift as a sample and hold will.

Note that the output is relatively high impedance. If you have a rail to rail op-amp, you could use this to buffer the output voltage.

The C and R in the key circuit can be 0.1uF and 1MΩ.
The Cs and Rs in the delay circuit can be 0.1uF and 100kΩ.
R6 could be 100kΩ
The trimpots could be 10kΩ
The diodes can be 1N4148

sorry but the main problem is Ive already built the keyboard,
and there isnt enough room inside the synth enclosure for all them components,
the sample and hold may have accuracy issues but it doesnt require any modification of the keyboard and its tiny.
so the problems im having are:

-I need a normally open electronic switch ,
after posting this thread I discovered analog switches and solid state relays, they seem like they would work fine, and Im positive a mechanical relay would work

-I need a rail to rail Op amp with the same or lower input bias current as a TL074 (200 pA)
I suggested the AD822 which apparently has ''25 pA maximum input bias current''
If this wouldnt work I now think the TL074 could maybe work!
my reasoning: when I was messing around with noise toaster before this project i distinctly remember that the frequency pot (top left of the schematic,this is what im duplicating for every key) must be turned up at least 1/6 before anything became audible,
a few minutes ago I tested what voltage a potentiometer voltage divider would put out when turned 1/6 the way up, it was around 1.75V.
the TL074 input offset voltage is 1.5V and its output tops off at around 8.75V , and it sample and holds perfectly, so that 1.5-8.75V range would be fine

- and suitable schottky diodes for the CV (green diodes in the diagram^) to prevent too much of a voltage drop

I have polystyrene capacitors already and I think thats all.

EDIT:I took some photos of the synth today and posted them below with a description of whats going on.

 

[AFex54]

Ive finally got some pictures of the actual synth

First things first: There are 2 Noise Toasters !!!

The overview : the grey knobs are the keyboard frequency pots, 2 knobs for a black key, 1 for a white key,
the black keys are 'polyphonic' i.e they output 2 seperate CV for either NT, with a keypress you send 9V to 2 voltage dividers rather than 1, 1 switch 2 pots.
the white keys are normal, the same as in the diagram^, 1 CV goes to both NTs.

youll notice there are no switches for the NTs, instead the switches are replaced with 3.5mm jacks (Dont worry this was not my idea, a followed a detailed guide and it works 100%) making the noise toaster a 'semi-modular' synth, like this Korg MS20
with 2 noisetoaster PCBs this adds MASSIVE increase in functionality and possible variations of patches, but most importantly its allows full communication between the seperate PCBs.




The guts:
You'll notice the 2 Noise Toaster PCBs!
the little PCB to the far right is just an LFO and also a pulse generator that is used as a basic arpeggiator for gating the envelope generator (explained here within the first minute) .
to the far left mounted vertically to the top NT PCB is the little board reserved for the keyboard's S&H circuit, there would be 2 circuits for the polyphony, one for each Noise toaster .
youll also notice the top NT is missing components... well its the secondary NT, it is missing the audio amplflier since you only need one

you can see the the wires from the keys connecting to their pots,
and the NT hardware is mounted on the panel but they are not wired to the PCBs yet






The backside: at the far right is power input and beside it the power output, its just a splitter, they're wired tip to tip and ring to ring.
I have 1.5A supply and the NTs consumes a grand total of 30mA ,
So anyway Im building a simple sequencer right now for use with this synth and I plan is to be able to power it with the same supply plus any other analog music machines that i may be building in the future.
... other than that not much else going on back here




so yeah Ive managed to make this 10 times more complicated for you but I hope finally having a visualization of this mystery 'synth' will make up for that.

 

[(*steve*)]

OK, here's a simpler version. Not as good, but a lower component count.

You will need a sample and hold, and you will need to buffer the output of this circuit. Also, as you may have noticed, the sample and hold will drift.



This is practically what you've been trying to do.

I have simply replaced the diodes and the FET with 4066 elements which have resistance, but no other forward voltage drop.

The circuit at the top is not as critical bit it allows to you set a definite delay and gate time for the S&H which should yield better consistency than just using the signal from the summing diodes to control the S&H.

If you want the S&H to hold the "note" for any period, you better look up guard rings, use a very good capacitor, and a very high impedance op-amp.

 

[AFex54]

OK, here's a simpler version. Not as good, but a lower component count.

You will need a sample and hold, and you will need to buffer the output of this circuit. Also, as you may have noticed, the sample and hold will drift.

View attachment 21556

This is practically what you've been trying to do.

I have simply replaced the diodes and the FET with 4066 elements which have resistance, but no other forward voltage drop.

The circuit at the top is not as critical bit it allows to you set a definite delay and gate time for the S&H which should yield better consistency than just using the signal from the summing diodes to control the S&H.

If you want the S&H to hold the "note" for any period, you better look up guard rings, use a very good capacitor, and a very high impedance op-amp.

I think the only way this would fit is if I could get CMOS switches in single, transistor-type packages to mount on the pots themselves .
Do these exist?

the op amp for S&H im a bit worried about, there is good chance a suitable one isnt available.
rail to rail, very high impedence, very low input bias current

thanks btw , this circuit is good.

 

[(*steve*)]

You can get single cmos switches in SOT-235 package. Remember that these need power connected as well.

 

[AFex54]

You can get single cmos switches in SOT-235 package. Remember that these need power connected as well.

damn it, I forgot about power...
well I dont think It'll be possible to do this, there will be 10 14-pin packages for all those pots! they will need a PCB and I barely fit in that board for S&H above (I didnt mention that the space on the right side is taken up by the speaker)

Also above when you said ''the sample and hold will drift'' do you mean droop, like the slow decay of the sampled voltage?