sine wave current booster / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

sine wave current booster

first previous 2 3 4 5 next

Move Thread

LAN_403

Dave47

Mon Oct 06 2008, 11:55AM

Registered Member #84 Joined: Thu Feb 09 2006, 01:06PM
Location: Dallas, TX
Posts: 47

Hazmatt_(The Underdog) wrote ...

That's a sweet amp but I'm not sure how to get 40Vpk pk out of +- 15V supply unless we're talking two supplies?

I have some THS4141's that I could use to drive both of those amp units with, and I might be able to really push them to high voltage output by adding their outputs. Might be fun... might be a lot of smoke... have to order a few and see. heh.

It's 40Vpp differential. So, if each output is able to drive 20Vpp single ended, and you drive both amps 180 degrees out of phase, you get 20Vpp - (-20Vpp) = 40Vpp.

The THS4141 would be excellent at driving this and doing the single ended to differential conversion at the input while maintaining the ability to do DC.

David

Hazmatt_(The Underdog)

Mon Oct 06 2008, 07:01PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

Okay good, I have to work on that. Time to get more chips!! ;)

Sulaiman

Tue Oct 07 2008, 06:28PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Hazmat - I hope that the NTE88 in your diagram is drawn wrongly - should be collector to -Vsupply.

The circuit will be WORSE with schottky diodes!
The 1N4007 diodes are acting as a temperature-compensated zener/reference,
if you use schottky diodes you will get a lot of cross-over distortion.

Hazmatt_(The Underdog)

Tue Oct 07 2008, 07:01PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

Oh HAHAHAHHAHA! YEA .. the transistor got reversed! they're both collector to supply so please don't build that amp! It will fail on power on!

I thought for a bit about the diodes too, you're totally right. The diodes have to be the closest match to the internal diode to the transistor, so 1N4148's would be a much better choice. Schottky diodes with their low forward voltage would cause a big imbalance to the transistors and the output would be crossover distortion.

I have to think these things out more because of my lack of experience with any real type of amplifier, but I am getting somewhere.. just ..hopefully sooner then later. heh.

Hazmatt_(The Underdog)

Sun May 24 2009, 09:14PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

Okay, I know its been a thousand years since the start of this thread, but I just wanted to say for the record that if you've never designed an audio power amplifier before, its not that simple without decent information. And I KNOW its a double post, but I don't really care, I wanted Turkey9 to know that I've been working on this for a long time and I HAVE NOT FORGOTTEN about him!

I didn't have a decent working front end driver/current mirror until I got "Practical Circut Analysis of Amplifiers". I highly recommend this book for future reference and design. "High Power Audio Amplifier Construction Manual" is good in most respects, and I may have been doing some things vastly wrong, but I was having difficulty with the basic amplifier presented. I could not get a response from the active diode bias generator and the constant current sources seemed to be in cutoff despite the bias voltages applied.

That being said, I have been somewhat successful recently in implementing power amps from my little test rig that I have in the picture. I was hoping to have a symmetrical drive into the power BJT's from a preamplifier stage, but as I have learned you need a tremendous voltage swing to accomplish that, along with an enormous voltage offset.
For a 55V supply rail you need the source to float at 52V and be able to swing down 20V, which could be done with an op amp, but would likely cause a large amount of signal distortion, and is difficult to drive with such a high offset voltage.

So what I have here for the time being is what could be considered a high voltage amp, rather then a high current amp. I do not know its full specifications just yet because its still in pieces, but it can easily pin the 20W scale into a 20R load with 55V supply rails.
The meter has a 20KHz cutoff frequency response, so I cannot get a decent power reading past that, but hopefully I will be able to find the output power of this amp sometime soon. Now that I have a working voltage amplifer stage/current mirror, this amplifier will work probably up to 120KHz, vastly short of your requirement, but I wanted to see this first amp through because it was a struggle, and I didn't want to walk away from it just yet without some closure.

The schematic that I am providing has 2 current mirror stages and performs just as well as what my circuit is right now, a single current mirror with voltage feedback. The circuit here also has a diode bias, but I am implementing variable bias generator using a transistor right now. And I am not yet using the high frequency compensation feedback shown yet. This is mostly for theoretical modeling purposes to try to see what the upper frequency limit might be with applied feedback.

So far I have just been desperate for 2 things, Voltage gain and Power gain. I say voltage gain because I want high voltage out. Class B drives give you higher current out but you lose 0.7V voltage from input to output. This is a real problem for my application and a real limiting factor when you have 22V peak to start with, its not a whole lot to work with!

One of the difficulties that I am having as we can see here is the amp is diode biased for 2.8V due to 4 diode drops with the darlington pair. This follows logically that you need to offset that voltage drop of the internal diodes of the transistors to eliminate crossover distortion. Well the problem with that is that my amplifier is vastly overbiased when using 4 diodes! It's burning up huge amounts of power into heat and is deeply biased into class A territory, when we want class AB or close to class B for 50% efficiency!
Therefore I am implementing the active diode bias and have found that ~ 1.56 volts works for now, about half the 2.8V bias, which I haven't figured out quite why it likes half Vbias, but I know it works okay. This may change in the future depending on how I can push the amplifier and how it behaives.

At this time I am using 2x NTE 87 and NTE 88 because I have not blown them up yet, heh. They are real troopers when it comes to power handling. However, they have a very poor frequency response and I will be changing over to NTE 248 and 285, or maybe if I can find an old T03 package that goes to 20MHz I will get those for this amp.
Overall the frequency response for this design will be poor compared to another mosfet amp design I have on hand which will come later ( I needed closure on this rediculous thing ). The frequency response tapers off because of Cjb and the current mirror. I wanted to force drive the amplifier with op amps or a class A stage to provide lots of current, but I could never get enough voltage gain and current gain to really turn on the sluggish BJT's. It is much easier with Mosfets, but like I say, I wanted to close the chapter on this amp.

I may use the flat T03P's in the future with a large heatsink, but the trouble is I have more heatsinks for the old T03's then the large flat ones, so its a matter of cost.

Amps of this nature, even with the newer TO3P's will likely not exceed 500KHz in frequency response even with a transition frequency of 500KHz, they just require too much current to be driven over that edge, but we'll see. After the completion of this sluggish power amp I will try to get 200W or maybe 300W out of the TO3P's and see how that goes. That is a matter of serious heatskinking though and I'm not sure my big heatsinks are capable of that task.

For now let's see how far we can get and with some practice I may get some decent amplifiers working.

Some bad news though bud, If you want serious power from 10KHz to 5MHz there is only one way to really do that, and that is with real RF transistors, the $80 each variety! That also requires transformer coupling and careful impedance matching. I recently bought a manual for the ENI 1040L which is a 400W 10KHz to a few MHz amplifier, and its not a real simple thing. I will have to be careful about the size of the ferrite beads I get so that they handle the power, but I will work on these soon. I want some real power amps too because I like toys too.

Things to come:
1. Mosfet power amp 10KHz to 700KHz: Talked to an amplifier company and they are helping me with this
2. RF transistor amplifier 10KHz to couple MHz, hopefully 20W, then higher power model.

Turkey9

Mon May 25 2009, 05:23AM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

Wow great job. I never imagined that a project like this would be so tough! I'm also glad that there was someone out there looking for the same kind of thing when I started this thread! Thanks for sharing all your hard work. This is still pretty high level stuff for me but I can follow it enough to replicate what you're talking about. When you get the final high frequency amp built I'm definitely going to try it!

Hazmatt_(The Underdog)

Mon May 25 2009, 07:10AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

I hope you do. I'm going to keep up the good fight as long as I have energy, heh. I would like to see several amps "captured" for forum reference so if someone needs a high power amp in some strange frequency range, they can take the design off this site and use it and have it work the first time.

Maybe we could even start a "Lab" tab for equipment so all the DIY'ers can copy and build their labs up, wouldn't that be cool!

Steve Conner

Mon May 25 2009, 09:26AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hey Matt, I see you got the John Linsley Hood book. So now you have to get Douglas Self's "Audio Power Amplifier Design Handbook" too!

IMO, it's considerably better at explaining how all those current sources and active diodes work than the JLH book. (They do work, honest!)

Have you actually seen the schematic of the Hitachi/Maplin MOSFET amp? It's about the simplest amp you can make.

It relies on the special properties of lateral MOSFETs to some extent, you can't just sub in an IRFP250 and an IRFP9250.

Actually I lied. It's the second simplest! Susan Parker's "Zeus" is the simplest. Maybe you could make an RF version of this with RF transformers instead of the big iron lumps she used.

Hazmatt_(The Underdog)

Tue May 26 2009, 03:09AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

I have a "Susan Parker" model RF amp and I would post a picture of it but my freggin camera batter dies in 3 minutes! gotta get a new one if they're still made.
The trouble with Mosfet push-pull RF amps is that at least with what I am using they have to be tuned to an extent, which undermines the entire purpose of being broadband. Also, very extremely! critical is the transformer! I have just a hair more experence then zero and the core is soo critical to the success of the amplifier!, I cannot express that enough.
For broadband amplifiers I urge everyone to use some kind of binocular type transformer because there is no way you will have much luck with the toroid in a serious power application.

The other trouble with the Mosfet push-pull amp is that it almost always requires filtering. Unfortunately there are a lot of harmonics that this type of amp generates and again undermines the purpose of being broadband because of output filter requirements.

What you posted is great Steve! and I know the books are okay, its just difficult when trying to implement and you get square waves instead of pretty sine waves. It may be a matter of overbias or underbias, or current imbalance, but honestly I just want to get something started and working.

I have more books on order right now. I'm unfortunately at a disadvantage because I'm a controls person and not an analog designer, so analog design is not a strength of mine, but I'm working on that.

I have lateral mosfets lined up, and a valid design to guide me. I have not yet used them because they cost a fortune. I may experiment with 2 of them and see what comes. The design I have on hand uses a class B driver as the signal source for the L Mos amp, so I need to get in there and validate that my class B driver is going to do the job, then see what transpires.

I don't want to post the exact design because I do not have permission from the amp designer to do that, but I will make mine available as soon as I can.

3 bookcases full.. starting on 4.

Steve Conner

Tue May 26 2009, 09:32AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hey Matt

When it comes to ferrite cores, there are two basic kinds of ferrite, manganese-zinc and nickel-zinc. One is for SMPS applications in the hundreds of kHz as well as EMI suppression, the other is for RF right up to hundreds of MHz. I can never remember which is which, but if you had trouble with your transformer, you probably got the wrong one.

The binocular cores are made specially for RF transformers, so they're guaranteed to be made of the "right" stuff. You can make good broadband transformers with them: my HF rig has a 100W RF amp that covers 1.8 to 30MHz using one of these. They do need to be tuned somewhat, but it's almost a case of "anti-tuning" to flatten out resonances and broaden the bandwidth.

first previous 2 3 4 5 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob