latfets

What people are working on at the moment
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#1 latfets

Post by ed »

I found 2 of these on the shelf, unopened from an abandoned GM70 project from way back.
smps.jpg
I thought I’d see what this sounds like, powered by aforementioned power supplies. Circuit shows 2sk1058 but the ecx latfet is supposed to be a drop in replacement and I couldn’t find a spice model for the ecx. It sims at about 3.5 watts with a bias at 1.6 amps and it’s very slightly down at 30hz. I’m keen to compare it with the 3.5 watts of 2A3 in the studio system.
ecx10n20.jpg
The vas on it's own sims well with a square wave, but the square wave through the whole circuit shows some downward drift over time. Not sure what that means at this stage but it's all good fun finding out. It's been interesting playing with the opamps in audio mode, chasing the ringing and getting faster rise times etc...

It needs the VAS cos the highest output source I have is 2v RMS and the fet can stand a lot more than that at this op point.

onward

btw: I put it here cos it's a project and I reckon nobody reads the SS section anyway...snigger
diet: full credit to Michael Rothacher for the choke loading ideas though.

pps: I was hoping for some comment from those more worthy, particularly with respect to shunting any ac injected into the supply rail viaR5 and X1 cos it may influence the bias set by R2 and R3..?
There's nowhere you can be that isn't where you're meant to be
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#2 Re: latfets

Post by Nick »

Cant help thinking a diode and a cap would isolate the front end better.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#3 Re: latfets

Post by ed »

I'm pretty sure you're right but it's on the edge of my comprehension at the moment. I'm currently reading ways of decoupling the bias on opamps, most of which involve a zener......

I'd be grateful for an illustration though.

The big problem as I see it is running the opamp single ended in the first place.......but what is life without these little challenges!

ah ha!, did you mean a diode in the R5 bias line?
There's nowhere you can be that isn't where you're meant to be
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#4 Re: latfets

Post by ed »

comme ca?
ecx10n20 se.jpg
There's nowhere you can be that isn't where you're meant to be
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#5 Re: latfets

Post by Nick »

No, a diode between the positive rail an the top of r3, then a cap from the top of r3 to ground. You could use teh same point to provide the +ve supply for the op amp, but that will have better psrr so less important.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#6 Re: latfets

Post by IslandPink »

Latfet Railroad ?
"Once you find out ... the Circumstances ; then you can go out"
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#7 Re: latfets

Post by ed »

more shenanigins
ecx10n20 se.jpg
this is really taxing my dementia now......I think:
the network R2/3 in parallel with C8 sets the bandwidth -3db position at 32hz which I think means anything below 32hz wont get shunted but could influence the power rail....oh waily...but better than the previous attempt....

I'm still scratching my head about ac leakage through R5 though

onwards
There's nowhere you can be that isn't where you're meant to be
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#8 Re: latfets

Post by Nick »

I'd be grateful for an illustration though.
Email me the .asc so I don;t have to draw it all out.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#9 Re: latfets

Post by Nick »

This sort of thing
Attachments
Amp.png
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#10 Re: latfets

Post by ed »

sorted, many thanks, now to round up the missing bits
There's nowhere you can be that isn't where you're meant to be
User avatar
jack
Thermionic Monk Status
Posts: 5493
Joined: Wed Dec 29, 2010 8:58 pm
Location: ɐılɐɹʇsnɐ oʇ ƃuıʌoɯ ƃuıɹǝpısuoɔ
Contact:

#11 Re: latfets

Post by jack »

There are a number of things about this circuit which are worthy of comment.

R2 and R3 set the DC operating point to Vcc/2 as is standard in any single rail non-inverting opamp design. Unfortunately, any PSU noise is also amplified as it's also presented to the non-inverting input, hence the addition of C8. However C8 would be an AC shunt to the input signal, so an attempt to isolate it is made by the addition of R10 (which also helps in balancing the currents into the inputs of the opamp). This approach is a compromise - also, the bandwidth is not set by these components - they are concerned with setting the DC operating point of the opamp only. The rolloff at 32Hz you are referring to is for power supply noise rejection only, not the amplifier bandwidth. Note that Vcc is is direct feedback path in these scenarios - this can lead to instability (oscillation at low frequencies, e.g motorboating).

The DC gain of the opamp is unity as C5 isolates R1, thus the DC output of the opamp is also Vcc/2, which we want.

At AF frequencies, the AC gain is 1 + R9/R1, i.e. 3, assuming C5 is big enough to have low Z at AF compared with R1. C7 provides a bit of loop stability by providing HF gain rolloff, reducing the gain by effectively lowering the value of R9 as frequency increases.

A better approach is to provide a bias which is isolated from Vcc so these compromises are avoided and the PSRR and noise is improved.

Lots of app notes out there about these issues, e.g http://www.analog.com/en/analog-dialogu ... blems.html which covers most of these points. Odd that the component values in the above app note (which I chose at random from those available) are identical to the schematic you have.
Vivitur ingenio, caetera mortis erunt
User avatar
jack
Thermionic Monk Status
Posts: 5493
Joined: Wed Dec 29, 2010 8:58 pm
Location: ɐılɐɹʇsnɐ oʇ ƃuıʌoɯ ƃuıɹǝpısuoɔ
Contact:

#12 Re: latfets

Post by jack »

Talking to myself wrote:Lots of app notes out there about these issues, e.g http://www.analog.com/en/analog-dialogu ... blems.html which covers most of these points.
Just seen that that app note was published as an EDN Design Feature aticle at https://www.ieee.li/pdf/essay/single_su ... design.pdf

Figure 4 shows the best of the simple designs... and, yes, you have positive feedback through X1 & R5 via Vcc, which is maybe why you had the DC drift,
Vivitur ingenio, caetera mortis erunt
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#13 Re: latfets

Post by Nick »

I might be tempted to use a tl431 and make it adjustable. If it wasn’t for the choke load which I guess could allow the source to go -ve, I would direct couple it and use the centre point to set the fet operating point. Or, you could use a DC:DC converter and generate a -ve supply for the op amp and most of the problems go away.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#14 Re: latfets

Post by ed »

jack wrote: Tue Jan 23, 2018 1:06 am
Lots of app notes out there about these issues, e.g http://www.analog.com/en/analog-dialogu ... blems.html which covers most of these points. Odd that the component values in the above app note (which I chose at random from those available) are identical to the schematic you have.
why is that odd?, that's exactly where I got the info from in the first place.
jack wrote:Figure 4 shows the best of the simple designs... and, yes, you have positive feedback through X1 & R5 via Vcc, which is maybe why you had the DC drift,
yes, fig 4 is exactly what I was referring to in my post earlier when I was talking about zeners..I haven't fully got my head around that at the moment. The feedback through X1 and R5 is still keeping me awake at night.

I do feel like I'm looking for excuses for going off at half cock on this but it's just that I was trying to address 3 of my previous abandoned projects in one go, i.e. investigate the sound of smps from F6 experiments, find out what latfets sound like, and try choke loading, which I abandoned when investigating SITs. I was maybe a bit ambitious here because the guys playing with Michael Rothacher's mofo were most often driving it with a high gain pre amp, something I don't have....
Last edited by ed on Tue Jan 23, 2018 9:38 am, edited 1 time in total.
There's nowhere you can be that isn't where you're meant to be
User avatar
jack
Thermionic Monk Status
Posts: 5493
Joined: Wed Dec 29, 2010 8:58 pm
Location: ɐılɐɹʇsnɐ oʇ ƃuıʌoɯ ƃuıɹǝpısuoɔ
Contact:

#15 Re: latfets

Post by jack »

ed wrote: Tue Jan 23, 2018 8:57 am
jack wrote: Tue Jan 23, 2018 1:06 am
Lots of app notes out there about these issues, e.g http://www.analog.com/en/analog-dialogu ... blems.html which covers most of these points. Odd that the component values in the above app note (which I chose at random from those available) are identical to the schematic you have.
why is that odd?, that's exactly where I got the info from in the first place.
Odd as in the odds were against it considering the number of app notes out there :)

Single rail non-inverting opamps are always a pain due to the biasing issue. I'm with the other Nick - use a voltage reference as per figure 4 and maybe replace the zener with a TL431. Zeners are less accurate and drift with temperature, so setting and maintaining the DC conditions can be tough.

Moving to dual rail gets rid of the biasing issue for the opamp but then you'll need a DC offset for the FET!
Vivitur ingenio, caetera mortis erunt
Post Reply