Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Instruments and Amps Everything that makes music, Especially including instrument amps.

Tube Emulation & EQ
Tube Emulation & EQ
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 16th February 2018, 05:18 AM   #351
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
Default Balanced line out - tips?

The guitar amp I've built for my friend is, like all engineering designs, a set of compromises. In order to keep it light (for my friend) and affordable (for me), I used boom-box speakers, which will limit the maximum SPL available.

I think it'll be loud enough for my friend 99% of the time, but for that last 1%, I'm thinking of adding a balanced XLR output. If he ever wants more loudness than the amp can provide on its own, he can use that to plug into the P.A.

Since the de-nastifying filter already has appropriate EQ / speaker emulation in it, this should be a pretty simple matter, i.e. I just need to tap off the signal at the output of the de-nastifying filter, and send that to the XLR jack.

However, this is one of those things I've never actually designed or built before. I want to avoid ground-loop issues, and make sure the signal level is in the right ballpark to plug into a microphone input of a mixer.

Anyone have any tips on how to do it? Should I use a signal transformer? If so, what kind? What sort of circuitry does it take to drive said signal transformer (i.e. expected primary impedance and signal level?)

I'm attaching a screenshot taken from a Rane document on "sound system interconnection" ( Sound System Interconnection ). Does this look like the way to go?

Thanks in advance,

-Gnobuddy
Attached Images
File Type: png unbal_to_bal_XLR.png (62.1 KB, 75 views)
  Reply With Quote
Old 17th February 2018, 05:35 PM   #352
mjd_tech is offline mjd_tech  United States
diyAudio Member
 
Join Date: Apr 2002
Location: Boston, MA
I've made an initial prototype of KMG LND150 gain stage, with fixed bias and Leo Fender cathode resistor (1k5). Actual components, not simulation.

For reference:
http://milas.spb.ru/~kmg/files/proje...m800mvfSch.pdf
and see attached schematic for my changes.

Power Supply: I discovered an extra 1:1 isolation transformer in one of my parts boxes, so I made a half-wave voltage doubler. For now, I use a variac so I can adjust the output to 300VDC under load

Parts substitutions: For the parallel BAV70 pair, I use a single 1N914, for the BAT54S schottkys I use through hole BAT85S, For the ES1J high voltage fast recovery diode, I use UF4007.

Fixed Bias: Instead of referencing the LND150 gate to ground, it goes to a 270k/2k7 voltage divider. The LND150 source circuit is referenced to ground instead of a negative supply. This setup requires capacitor coupling the input signal, so I use 22n.

DC operating point:
  • Drain: 196V This is close to a Leo Fender stage where the plate sits at roughly 2/3 of B+. So we're getting about 1mA current through the 100k Drain resistor.
  • Gate: 2.71V
  • Source: 3.38V
    This gives a Vgs of -0.67V. This is reasonably close to what you would get drawing a load line with the LND150 "output characterists" curves.
  • Top of Source diode chain: 3.20v
  • 1k5 resistor: 1.48v
    So the diode chain drops 1.72v @ 1mA. This seems reasonable for 4 schottkys and one regular silicon diode.
AC characteristics:
I fed a 400hz sine wave at various voltage levels, and monitored output with a scope. I measured gain with a 100mV peak-to-peak input signal, so there would be no clipping.
  • With the 1k5 source resistor unbypassed, the gain was about 32.
  • Fully bypassed, the gain was about 80.
  • Bypassed with a 470R resistor in series with the bypass cap, gain of 58.

Note: my scope doesn't have fancy digital readouts, so gain measurements are by eyeball, so consider them "ballpark" figures. I do all AC measurements by peak-to-peak values on my scope, it's easier that way.

So here, LTspice is a little optimistic with gain, versus a real circuit.
LTspice said a 910R bypass resistor for a gain of 60. Reality said a 470R resistor for a gain of 58.
So this resistor is "adjust on test".

Next, I gradually increased the input signal to overdrive the gain stage. From what I could tell, with my less than stellar scope, here's what happens:
  • As the input signal increases from 100mV, the output looks like a sine wave and gets bigger, as expected.
  • At around 1Vpp input, the bottom of the output waveform starts looking less sine wave like, a little "peakier". It's still rounded, and it will stay rounded even under massive overdrive. This appears to be the threshold of "grid current".
  • Around this time, the top of the output waveform increases a little less than the bottom.
  • At around 3Vpp input, the top of the waveform increases no further. It is technically clipping at this point, but the leading and trailing edges are still very rounded. Bottom of waveform still rounded, and now increases slower.
  • At 4Vpp input, obvious clipping on the top, bottom is still rounded. The duty cycle is changing.
  • At 6Vpp input, the duty cycle changes even more, the top clipping is more pronounced, but still with slighly rounded leading and trailing edges. See attached photo. Pardon the crappy photo, but you get the idea.
  • At 12Vpp input, the trend continues, more (or less?) duty cycle, bottom still rounded.

For giggles, I put a 220R in place of the 180R hanging off the Source. It changed the DC voltages slightly, but didn't seem to make a difference in the AC characteristics. Again, my scope is not the most accurate thing in this world.
I'm sure the "grid current" threshold does change slightly, but not radically.
I couldn't tell the difference. The output waveforms behaved the same as with the 180R.
KMG uses 180R, 200R and 270R in his preamp. I'd say 220R is ok too.

I have not yet applied an actual guitar signal to this and listened to it. I don't expect just one stage will have any "magic", it needs a complete preamp. Just this one stage used up most of my little breadboard, it's a fragile mess, so I need to build on perfboard next.

Overall, the experiment worked out, at least from a DC operating point and sine wave perspective. No smoke, no surprises. Pretty darn close to what a 12AX7 would do in similar circumstances.

The question remains whether the 270k/2k7 voltage divider can be used for multiple gain stages, or if it needs a separate divider for each gain stage. Probably needs a separate one for each. We are after all simulating "grid current" here and the current has to come from somewhere. So connecting multiple gain stages to a single divider might be a recipe for oscillation, or not? It could do something really cool or not make any difference?
Attached Images
File Type: png LND150-Fender-proto.png (31.8 KB, 69 views)
File Type: jpg LND150-6VPP.jpg (118.9 KB, 66 views)
  Reply With Quote
Old 17th February 2018, 09:17 PM   #353
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
Quote:
Originally Posted by mjd_tech View Post
for the BAT54S schottkys I use through hole BAT85S
I wonder if a single 1N914 could replace a pair of BAT85S?

It seems wasteful to put four Schottky diodes in series if there is some other kind of diode that produces about the same voltage drop and dynamic resistance (they all seem to follow the same exponential curve at these low currents, so they really should have the same dynamic resistance if they have the same voltage drop at the same current.)

Maybe one LED, and one UF4007 in series for high-voltage protection?

Quote:
Originally Posted by mjd_tech View Post
For the ES1J high voltage fast recovery diode, I use UF4007.
Sounds good! You just reminded me that I totally forgot to buy any fast, high-voltage diodes when I bought Schottky diodes to try KMG's ideas out. Drat!

Quote:
Originally Posted by mjd_tech View Post
So here, LTspice is a little optimistic with gain, versus a real circuit.
I found the LTSpice model for the MPF102 JFET to be way off from reality. It seems you're finding the same thing for the LND150 model.

Supertex provides an LND150 model, but the one I posted in this thread was tweaked by someone who found the factory model was a very poor fit to the published LND150 datasheet curves. But it seems it doesn't match your actual LND150 devices very well, either.

I wish I knew how to tweak LTSpice models to better match reality. It can't be that hard, but I haven't stumbled across any documentation on how those diode and FET models are created. Staring at the text inside existing LTSpice models hasn't been enough for me to figure out how they work.

Quote:
Originally Posted by mjd_tech View Post
Just this one stage used up most of my little breadboard
Now that's a little annoying! There are a lot of components per stage, aren't there?

And once again I'm wondering if four Schottky in series can't be replaced by some smaller number of components.

Quote:
Originally Posted by mjd_tech View Post
The question remains whether the 270k/2k7 voltage divider can be used for multiple gain stages, or if it needs a separate divider for each gain stage.
My guess is that you can run at least a few stages off the same divider, especially if you increase the 100nF filter cap to maybe 10 uF or 100 uF instead. (Make it big enough, and you can simulate the heater warm-up time and slow turn-on of real valves!)

I wonder if replacing the 2.7k with an appropriate LED might also be worth a try? The lower dynamic resistance of an LED (compared to a 2.7k resistor) might help with running more stages off the same voltage node.

Quote:
Originally Posted by mjd_tech View Post
We are after all simulating "grid current" here and the current has to come from somewhere.
True, but you have a 1 meg resistor between the FET gate and the 3V DC node, so you can't draw more than a few microamps of current from your 3V DC source.

That's why I suspect you can feed a number of stages from that same point, as long as the dynamic impedance to ground is low enough not to create any unwanted feedback.

A 100nF cap has a rather large 20 kilo-ohm reactance at 80 Hz (guitar low E), so it really isn't helping filter that node much - the 2.7k resistor is a lower impedance than the capacitor over most of the guitar frequency range.

But a 100uF cap will get that reactance down to 20 ohms, which I think should be negligible in that circuit.

Thanks for sharing your results so far! I look forward to reading about your future developments - it will be interesting to see if KMGs ideas work to create good Fender Blackface clean tones.

I think the clean tones I liked from KMGs FET Bogner Ecstasy were actually from his 2SK216 MOSFET amp, rather than his later LND150 one. The circuit was a bit different too ( http://milas.spb.ru/~kmg/files/proje...fet/XtcFet.pdf ), with one source diode and apparently no grid-current emulation.

So I'm very curious if his later LND150 circuitry provides equally good (or better) clean tones.

-Gnobuddy
  Reply With Quote
Old 18th February 2018, 01:51 PM   #354
Bigun is offline Bigun  Canada
diyAudio Member
 
Bigun's Avatar
 
Join Date: Jan 2009
Location: Waterloo, ON or Herefordshire
Tube Emulation & EQ
Quote:
Originally Posted by Gnobuddy View Post
So I'm very curious if his later LND150 circuitry provides equally good (or better) clean tones.

-Gnobuddy
If you are looking for low distortion for 'clean' then with most FETs (all ?) the linearity will improve with standing current. If you can keep the little thing cool then increasing current will buy more linearity. You can also fine-tune the source degeneration resistor for linearity (load line). But for a guitar amp, I really doubt it's going to be audible once you have the device operating at a reasonable current level to start with.
__________________
"The test of the machine is the satisfaction it gives you. There isn't any other test. If the machine produces tranquility it's right. If it disturbs you it's wrong until either the machine or your mind is changed." Robert M Pirsig.
  Reply With Quote
Old 18th February 2018, 06:34 PM   #355
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
Quote:
Originally Posted by Bigun View Post
If you are looking for low distortion for 'clean' then with most FETs (all ?) the linearity will improve with standing current.
The trouble is that the FET circuits I've tried have all been too linear, so they sound too clean (tending towards lifeless and sterile, like so many solid-state guitar amps.)

It seems we have to do something to make them less linear, in a good way, if we want to come close to what valves do naturally.

I estimated 35% second harmonic distortion for the good-sounding "clean tones" from my DIY valve preamp with a 6AG5 final preamp stage ('scope screen capture was posted earlier.) I have been having trouble squeezing anywhere near 35% out of a JFET.

I found some posts by KMG in another forum, and he provided a link to the following clip, with the statement that it came from his MOSFET version of the Bogner Ecstacy: http://milas.spb.ru/~kmg/files/proje...mg_xtc_mix.mp3

Those sound like pretty decent guitar clean tones to me. I think the circuit that created those "clean tones" (which are not clean at all in Hi-Fi terms, but incorporate enough distortion to sound good for guitar) is this one, using 2SK216 MOSFETs and an earlier evolution of his triode-emulation circuit: http://milas.spb.ru/~kmg/files/proje...fet/XtcFet.pdf

I don't fully understand this earlier form of MOSFET triode emulation either, but I think the diodes connected to the source of each MOSFET are there to lower transconductance and add nonlinearity, to roughly match what half a 12AX7 does.

-Gnobuddy
  Reply With Quote
Old 18th February 2018, 10:21 PM   #356
mjd_tech is offline mjd_tech  United States
diyAudio Member
 
Join Date: Apr 2002
Location: Boston, MA
Quote:
Originally Posted by Gnobuddy View Post
I wonder if a single 1N914 could replace a pair of BAT85S?
I tried that. Replaced 4 schottkys with 2 1N914.
When overdriven, the leading and trailing edges on the top of the waveform are sharper, still not completely squared off, but not as soft as with the schottkys.
However for moderate signal levels, say 100mVPP to 1VPP, it's hard to tell the difference. The schottkys (all 4 of them) seem to be critical for simulating the 12AX7 characteristic curves that get closer together near cutoff.

Quote:
Originally Posted by Gnobuddy View Post
I found the LTSpice model for the MPF102 JFET to be way off from reality. It seems you're finding the same thing for the LND150 model.
This isn't the first time I have found spice models to be off
Quote:
Originally Posted by Gnobuddy View Post
My guess is that you can run at least a few stages off the same divider...
I wonder if replacing the 2.7k with an appropriate LED might also be worth a try? The lower dynamic resistance of an LED (compared to a 2.7k resistor) might help with running more stages off the same voltage node.
Well I didn't have any LED's that would work out for the required voltage. I have some red ones at 1.6v and some green at 1.8v.

What does work is 4 x 1N914, gives around 2.8 volts, which remains fairly constant with fluctuations in B+ (I turn the variac up and down).

For my version of the circuit (different diodes than KMG and fixed bias), it looks like the sweet spot for the bias supply is around 2.75 volts give or take a tenth.

There are through hole 2.7v zeners, but I don't have any, but I do have a bag of 1N914 so will probably go with those for now.

Quote:
Originally Posted by Gnobuddy View Post
I think the clean tones I liked from KMGs FET Bogner Ecstasy were actually from his 2SK216 MOSFET amp, rather than his later LND150 one. The circuit was a bit different too ( http://milas.spb.ru/~kmg/files/proje...fet/XtcFet.pdf ), with one source diode and apparently no grid-current emulation.

So I'm very curious if his later LND150 circuitry provides equally good (or better) clean tones.
You and me both. I've still got a lot of work to do with this, but the important first steps have been taken.
  Reply With Quote
Old 18th February 2018, 10:52 PM   #357
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
Quote:
Originally Posted by mjd_tech View Post
The schottkys (all 4 of them) seem to be critical for simulating the 12AX7 characteristic curves that get closer together near cutoff.
Ah. Thanks for trying the experiment, and telling us how it turned out. Okay, four Schottky's it is!

-Gnobuddy
  Reply With Quote
Old 19th February 2018, 06:07 AM   #358
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
It just occurred to me that an LND150 with a 100k anode resistor will probably have an output impedance of essentially 100 kilo ohms. But half a 12AX7 with a 100k anode resistor will have an output impedance of around 40 kilo ohms, two and a half times lower.

This is because the (internal) anode resistance of the triode is somewhere in the ballpark of 70 kilo ohms, much lower than the internal drain resistance of a modern MOSFET.

In other words, we can expect considerably more severe loading effects on one of KMG's triode-emulation stages, compared to the real thing.

-Gnobuddy
  Reply With Quote
Old 22nd February 2018, 03:50 AM   #359
Gnobuddy is offline Gnobuddy  Canada
diyAudio Member
 
Join Date: Mar 2016
I added a line-out jack (actually a mic-level XLR male connector) to the amp, so my friend can plug it into the P.A. if he ever wants more volume than the pair of 6 1/2" speakers can put out.

This was pretty easy to do because the de-nastifying filter already removes all the nasty-sounding frequencies before they go to the internal class-D power amp module. So all I had to do was tap off the signal from the de-nastifying filter, pad it down to the right level, and add transformer isolation to prevent ground loop problems when plugging into the P.A.

It turns out Triad makes an isolation transformer with a full 20 Hz - 20 kHz bandwidth, and it works at impedance levels that are comfortable for both the de-nastifying filter and a typical microphone input on a mixer. It's affordable too, not like the platinum-and-diamonds Jensen mic transformers. (TY-250P Triad Magnetics | Mouser Canada )

As you can see in the schematic, I wired a 1k resistor across the transformer primary, and another across the secondary. A glance at the attached transformer frequency response plot will make it obvious why I added the 1k resistors!

We tested the line-out at last night's jam, and it works very well. My wife commented that she had never heard our friend's playing as clearly as she did last night.

I think the amp is pretty much done now, and none too soon. I'm burned out from having spent all my spare time on this project for some months now.

I will get some photos of the completed amp next week, just to wrap up the story here.

-Gnobuddy
Attached Images
File Type: png lineout_2018-02-21.png (27.7 KB, 24 views)
File Type: png TY250-P_Frequency_Response_2018-02-21.png (45.4 KB, 24 views)
  Reply With Quote

Reply


Tube Emulation & EQHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
CA 3080/LM 3886 tube pwr amp emulation metalhead Solid State 2 3rd February 2018 04:10 PM
SIT Device Emulation poynt99 Pass Labs 1 27th April 2014 10:14 AM
Tube emulation with transistors fireworks Tubes / Valves 17 4th February 2014 07:03 PM
Tube emulation (pritchard article) Barre Instruments and Amps 7 4th November 2010 02:28 PM
Tube amp semiconductor (FET?) emulation danV Instruments and Amps 11 24th June 2008 10:40 PM


New To Site? Need Help?

All times are GMT. The time now is 06:19 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki