Complementary pairs have their use, for example in output stages, or specific low-noise preamps, but you're mentioning VCA's. The Blackmer VCA uses complementary pairs, but your example shows a basic OTA, using a Long Tail Pair, which does not use complementary pairs. On the contrary it uses identical transistors.
Dual MOSFET in a package.
- Thread starter Analog_Fan
- Start date
Help Support GroupDIY Audio Forum:
They are both very good for medium power switching applications. They are not designed for audio, but you may find applications where they excel.
Actually, the lowest noise BJT's (ZTX851/951) commonly available are medium power transistors, where low-noise is just a consequence of the large geometry.
The manufacturer (Diodes Inc.) does not even mention they are low noise. That shows the audio industry is not a target anymore for industries.
user 37518
Well-known member
I don't know what you mean when you say that "FETs don't seem to cut the mustard". In what way they don't work properly? May I suggest you to download the models from Linear Systems for the 389
https://www.linearsystems.com/produ...=products&partnumber=LSK389 Series&match=true
The models don't seem to be super fancy, but try them. Which simulator are you using? if you are using the same Java-script online simulator you have shown in your other posts, then, I would say that you are not doing things properly. I used to use that simulator to teach my students about the basics of electronics, not to simulate more advanced circuits like the one you are trying to make. You should use LTSpice or something more suitable for the task.
user 37518
Well-known member
He has yet to tell us what exactly is wrong with the schematic or why things are not working for him. Everything he has said so far is extremely vague, I am not accusing him, but so far, all I have read is him saying that JFETs don't seem to be the answer so he wants to use MOSFETs, without telling us exactly what is going on, what he is looking for, or why he is inclined to that decision.
Analog_Fan
Well-known member
the Falstad simulator
Analog_Fan
Well-known member
Wel, the falstad simulator showed that if i use jfet's the "audio" aint centered around zero, but above.
user 37518
Well-known member
The falstad simulator is a "learn electronics" java simulator, like I said, its not a proper simulator, I used it to teach my students how transistors work and how the tiny current dots move, but thats about it. Get LTSpice, which is free BTW, or any other proper simulator before going further.
Analog_Fan
Well-known member
thnx for the suggestion, but a fews years back, i choose not to clutter my pc again.
i found this ic 2 days ago:
A (DVANCED) L (INEAR) D (EVICES)
ALD1105
what do you think of this?
what about pin 4 and 11 should they be treated like
like pin 13 on the CA3046, connect to a more negative/postive voltage or not use them at all?
user 37518
Well-known member
Don't take this the wrong way, but you are being rude. Abbey already told you that simulation models are very important, I already told you that the simulator you are using is not the proper tool to make the decisions you are making. Yet, you decide to dismiss everything, and then, you proceed to ask for more advice on something different, you insist on having it your way and want us to be a part of it.
Analog_Fan
Well-known member
Well, it seams i have to stick to parts i have chosen.
Mosfets, Can't find dual p-channel Jfets at mouser, it autonomic flips to mosfet.
Have contacted Paul from Falstad, several times, i trust him, you're be able to introduce the data of your parts.
precision isn't required, just wanna prevent buying that end up not being used after spending like 2 hours soldering.
i realized that the "ALD1105" has connected gates, not usable.
The gates are all independent. What are you talking about? Are you getting confused by the substrate connection brought out on a pin instead of connected internally?
Analog_Fan
Well-known member
i guess.
I'm using 5 volt, sourced from a Zener.
the current isn't very high, less than 1 mA.
that's the goal.
a 14 pin package is better than 2 x soic8.
I guess i should have looked a little deeper to page 5.
What about pin 4 and 11? ignore them?
When in doubt, follow the datasheet recommendation. The "Pin Configuration" diagram shows clearly pin 4 labeled as V-, pin 11 labeled as V+, meaning the substrate pins can connect to power supply voltages.
Many MOSFET devices connect substrate to source internally, but for that device with shared substrate you would only do that if using a circuit configuration which connects the source pins of both devices together (see the examples in the "Typical Applications" section of the datasheet showing current mirrors and differential pairs with substrate pin connected to source connection).
Lee_M
Well-known member
The datasheet from Sound Semiconductor (SSi) shows some options for reducing the noise of the 2164 VCAs by using smaller values for input and output resistors.
Using 7k5 instead of 30k apparently improves the noise figures by about 8-8.5dB, the 2164 VCAs probably compare a lot more favourably using those values.
(See the "performance" section on page.2)
https://www.soundsemiconductor.com/downloads/ssi2164datasheet.pdf
Rob Flinn
Well-known member
The figure that Porkyc measured was all 4 of the VCA's in the 2164 in parallel which reduces the noise considerably, but it's still noisier than a black dbx vca, which are about the most noisy vca's. However he was running class A which is supposed to be noisier than class AB
Last edited:
With VCA's there's always a trade-off between noise and THD.
"A 20kΩ value for RIN is recommended for most applications, but can range from 7.5kΩ to 100kΩ – lower values will produce the best noise
performance at some cost in distortion.
Maximum input current handling is approximately 1mA peak. This input current “headroom” is only likely to be a consideration when using
RIN values of 10kΩ and below with supplies of ±12V and higher. In such cases, one may want to design the signal chain for a maximum input
current of 900μA to allow adequate headroom."
+1 trading noise for distortion...
While not widely discussed the Blackmer VCA and sundry variants are "current ratio-ing" devices... The VCA input resistor feeding a virtual earth just converts the input voltage to an input current.... after current ratio-ing scales that input current up/down, another resistor converts that scaled output current back into a voltage. This is likely TMI too.
JR
neil.johnson
Well-known member
For the 2164, for low noise you want low current flow in the transistors so run it in class AB mode, the compromise being higher crossover distortion. For low distortion run it in class A mode, eliminating crossover distortion, but now with higher quiescent current the noise is higher.
One bandaid to this is sliding bias: at low signal levels (noise dominates) run it in class AB, then as the input amplitude rises (distortion dominates) slide the bias over to class A.
One bandaid to this is sliding bias: at low signal levels (noise dominates) run it in class AB, then as the input amplitude rises (distortion dominates) slide the bias over to class A.
moamps
Well-known member
I would just like to add that the N and P MOSFET pairs in the ALD1105 are not paired with each other (N with P) and that the difference is considerable. In addition, the P JFET pair LSJ689 is available, which can be paired with the N JFET pair LSK489 with some effort. The price is high (Part Search Results For "linear systems"), but the days when good JFET transistors were cheap are long gone.
Lee_M
Well-known member
The "4 sections in parallel" trick gives you 6dB noise reduction, so you'd get slightly better noise performance (up to 2.5dB better) by using lower value resistors (at the cost of THD), that should push the 2164 noise figures closer to those of the V2181.
Running in Class AB mode would reduce it further still (by up to 8.5dB, according to the datasheet), although again, at the cost of THD.
Using all 3 arrangements would be worthwhile if minimising noise is the main objective, but THD performance would be worse than porkyc's test results.
