More linear amplification

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
IanO, aside from my desires to achieve linear amp. using a FET transistor, I'd like to ask you: is the accepted method of amp. with a JFET to bias the gate along the transconductance curve somewhere between pinch off and zero volts Vgs. And voltage divider biasing of the gate should never be close to the pinch off voltage of the FET channel. Just enough to keep the signal from forward biasing the FET gate? If you'd like to share the equation for calculating the Siemens from data sheet zero gate volts, rated Vsd, and rated mAdc drain current, it would be much appreciated. My legal name is David E. Furlong, you might have heard the name Kevin Furlong; he's a brother who gave me this sites name.
 

Ian0

Joined Aug 7, 2020
9,667
Graphically, by drawing the load-line, is one of the most intuitive ways of designing the gain stage. This section from Blencowe Designing Tube Preamps for Guitar and Bass describes how to do it for a pentode valve, which has curves of much the same shape.
If the distance between intersections between the load-line and the I/V curves is constant, then the stage is linear.C863A164-3C44-432E-8B34-2DB1EA8053C1.jpeg9C1525DD-EA89-4BB8-97F6-09AE8160D4E1.jpeg1012DB73-931A-4938-BB74-3D80587EA164.jpeg
 

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
Graphically, by drawing the load-line, is one of the most intuitive ways of designing the gain stage. This section from Blencowe Designing Tube Preamps for Guitar and Bass describes how to do it for a pentode valve, which has curves of much the same shape.
If the distance between intersections between the load-line and the I/V curves is constant, then the stage is linear.View attachment 259064View attachment 259065View attachment 259066
Got it Ian, and it looks like for a certain anode resistance and signal input voltage range the operation of this pentode can become nearly linear. Next for me, is to apply that to my JFET and MOSFET broadcast FM tuner-decoder and preamp circuit design. I do intend to build up an amplifier for my tuner from discrete bipolar transistors and my "resistive coupling" mentioned in my "conclusion to an unfinished matter" (or whatever I called my post before I was suspended temporarily by the site). I doubt I'll ever get anything to function to any extent and I'll have learned my lesson about trying to be a one-man R&D company working on outdated technology, but maybe then I can let the matter rest.
Also, I think I've found the formula to use for drawing a load line and to calc gain/Siemens in FET circuits: Id = Idss*(1-Vgs/Vp)^2. A load line can hopefully be drawn from the FET's "ON CHARACTERISTICS" and numbers from that equation, before my decision to build up a circuit. It's something peaceful to do, can you relate?
 

Ian0

Joined Aug 7, 2020
9,667
2N5485 datasheet isn't one of the clearest. The On-Semi version consists mainly of Smith charts of its UHF properties: the Fairchild version is better for LF properties, but there are no V/I curves beyond 1 Volt.
However, there is a Id/Vgs curve, and you can see that there is a sweet-spot at Vgs = -0.75V where it looks quite linear.
 

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
2N5485 datasheet isn't one of the clearest. The On-Semi version consists mainly of Smith charts of its UHF properties: the Fairchild version is better for LF properties, but there are no V/I curves beyond 1 Volt.
However, there is a Id/Vgs curve, and you can see that there is a sweet-spot at Vgs = -0.75V where it looks quite linear.
Yeah, I'd like to see that I'd/Vgs curve, is it on the provided link. That voltage of 0.75V can be voltage divider biased into the gate, and I need an RF amp before a MOSFET preamp output. I noticed some data on reverse current of I guess the junction capacitance, that probably follows a charging pattern like a capacitor. And one of the matters I'd been trying to find a way of relating to you, has everything to do with the time constant of capacitor charging. The time constant for a cap in series with a resistor tends to set-up, if allowed, a repetitive waveform: a distortion from a true musical note waveform (I think). But anyway, a cap in shunt with the drain resistor, and of the same pF as the junction pF should take more of the non-linear jFET response in it's output. In otherwords, I guess I like that data sheet, but didn't find the I'd/Vgs curve. Is it on the attached data sheets?
 

Ian0

Joined Aug 7, 2020
9,667
It's the top left graph on page 3.
The graph you really need for designing with a ruler is this one, but it's not in the 2N5485 datasheet (this graph is just an example - it doesn't belong to the 2N5485)
2SK208.png
There are a few provisos:
1. The graphs shows maximim and minimum curves, both of which appear linear around that point. Presumably a "typical" device is also linear.
2. The 2N5485 is old. When the datasheet was draughted it was probably done by one bloke writing down numbers in a notebook, taking it to the drawing office, where another bloke, who probably wore a tweed suit and smoked a pipe, turned it into the datasheet using a ruler and French curves. The datasheet might not be as accurate as you might think.
3. The device has survived several company mergers. It might not be the same device any more. It may be an imposter. When two firms merge that are making two similar devices, then they tend to make just one and label it as both devices.

You need to bias the gate to MINUS 0.75V. You don't do that with a potential divider on the gate - you place a resistor between source and ground selected to drop 0.75V at the chosen drain current. Then put a capacitor across the resistor so that you don't lose gain.
The input capacitance is insignificant - specified as between 1pF and 3pF. Don't forget that this device is designed to work at 200MHz. You are working four orders of magnitude below that. You can forget about the gate capacitance, and putting 2pF across the output capacitance won't make any difference either. It will be less than circuit board capacitances.
 

Ian0

Joined Aug 7, 2020
9,667
I'm 6 miles east of the Greenwich meridian - mid afternoon on a cold sunny winter's day.
From here USA is between 5 hours (New York) and 8 hours (Los Angeles) behind.
 

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
I'm 6 miles east of the Greenwich meridian - mid afternoon on a cold sunny winter's day.
From here USA is between 5 hours (New York) and 8 hours (Los Angeles) behind.
IanO, please try to acknowledge receipt of my "private conversation" on this site before you call it a day online.

Mod: @Ian0

Use the'@' symbol and name to Alert a fellow user.
 

Ian0

Joined Aug 7, 2020
9,667
I can't: it says "This conversation is closed for new replies". Did you criticise a moderator? Don't do that!
 

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
I can't: it says "This conversation is closed for new replies". Did you criticise a moderator? Don't do that!
No Ian, I had remarked that I wanted to talk to you freely, and privately through this site because someone (Eric Gibbs) told me "stay on topic or else" after chatting with you about myself and brother Kevin. Yaakov told me to reach you on private conversation by clicking on your flag-emblem and I'd be on a private line where I figured I wouldn't need to stay on topic. It seems like we've got a little input from some guitar players. And I wanted to ask the device # of that FET with 2mA @.8Volts - and to be certain I've got the biasing correct, a 390 ohm resistor should put the gate at just under .8 volts - correct? And if you want, recommend a couple manufacturers. All of doing that simplifies the RF amp stage significantly. Please let me know if you get this message
 

Thread Starter

PatrickMalarkey

Joined Oct 2, 2021
120
Hi Ian,
No hang up at this end, perhaps Patrick has mis-clicked a box when creating the PM.??

E
You've got it Eric, there were two choices, 1 for adding anyone to the conversation and one to exclude outside comments. I thought for private messages choose exclude others but not Ian. Why did you tell Ian to hang up at this end?
 

Ian0

Joined Aug 7, 2020
9,667
No Ian, I had remarked that I wanted to talk to you freely, and privately through this site because someone (Eric Gibbs) told me "stay on topic or else" after chatting with you about myself and brother Kevin. Yaakov told me to reach you on private conversation by clicking on your flag-emblem and I'd be on a private line where I figured I wouldn't need to stay on topic. It seems like we've got a little input from some guitar players. And I wanted to ask the device # of that FET with 2mA @.8Volts - and to be certain I've got the biasing correct, a 390 ohm resistor should put the gate at just under .8 volts - correct? And if you want, recommend a couple manufacturers. All of doing that simplifies the RF amp stage significantly. Please let me know if you get this message
It's thanks to moderators like @ericgibbs that this forum doesn't degenerate into a slanging match like twitter, so best to keep to the rules.
We're now back on topic!

Do you just happen to have stock of the 2N5485? It's best as a UHF device, as its 1/f noise corner is rather high and its noise-figure is none-too-good. 2SK3557 is a better Audio FET and the datasheet has all the useful graphs.
Anyway, if you stick with the 2N5485, 390Ω is correct. You'll need 22μF in parallel to get maximum gain, or you can leave it out but get lower gain but maximum linearity.
What is it you are amplifying in the RF region? Don't ask me about RF amplification - it's really not my field!
 
Top