Leach amp

[destroyer X]

It is impossible to recognize.

When we construct some, our own self defensive system prepare us tu defend ourselves against frustrations.

Look that, if you spent some hundred bucks and some hundred minutes or hours constructing some equipment.... and if you deduce that it is awfull, you will be giving you a Degrée of Stupid....this way..... our inconscience... sub conscience.... treat to change whole things... and block this real analyses.

If you perceive a loose in bass, normally you will say that is because bass is precise!.... or because no harmonics, the sound is not so big and thunder because theres only fundamental frequencies on there.... or you think that others makes a damn mass in bass frequencies.... and the one you make is the only one on earth that is making good sounds.

Same way we cannot see that our children is bad!....cannot see our wife is awfull.... cannot see when we are stupid..... cannot perceive we were guilty and we are collecting whole we planted in past.

This way, i can not trust in the constructor opinnions, because may be modified by defensive systems.

How can a man that passes his entire life thinking that avoid sex is the best way to find God!.... and someday he discovered that his god do not exists..... look, the way he think god is....good!, and he perceive someday some bad things on his god....he will never believe in that!... he will think demon is taking his mind.... he will be defended, if not, he will see that he was, whole life, completely, no doubts, truly, with sure, a completely fool, idiot, ignorant and brainwashed!...who can support that without suicide... this way our system defend ourselves.

Ask americans if they are bad..... they will say NO!... WE FIGTH FOR WHOLE WORLD FREEDOM!.. DEMOCRACY FOR ALL... and money for them. Cannot see true.... TELL NORDISH THEY ARE COLD, THEY SAY NO!...YOU ARE CRAZY... TELL BRAZILIAN HIS COUNTRY IS A MASS, WE SAY NO!... WE ARE HAPPY! TELL JAPANESE THEY WORK TOO MUCH, WHEN YOU HAVE TIME TO YOURSELF?...THEY WILL ANSWER YOU, YOU ARE POOR BECAUSE YOU DO NOT WORK, YOU LAZY!

This way, other people evaluations, the ones that construct, evaluates in comparisons and keep them can be better, also the ones that construct and dismount because do not like are more precise, because no defended.

And the more crazy ones, that only likes really death of zen, are difficult to satisfy.(me)

Carlos

 

[djk]

"TELL BRAZILIAN HIS COUNTRY IS A MASS, WE SAY NO!... WE ARE HAPPY!"

My sister lived in Belim, Para, for 15 years.

A nice place to visit, but I wouldn't want to live there.

 

[djss]

I will agree with the statement that this amp is silent when idle. The only place I can hear it is within 2" of the direct center of my Morel MDT-33 tweeters.

A couple things to note from someone who's beat the hell out of these amps:

My 2-channel (1 channel when bridged) leach amp has a 1500VA toroid from Plitron, followed by 4x 37,000uF caps.

The amp hardly warms up at all when pushing an 8 ohm load.

Make sure your preamp can push the juice this amp needs. This amp needs a pretty high level input to really bring it to life. I plan to include a small preamp in the chassis with the amp for my new v4.5 amps.

- The amp WILL drive low impedence loads to full power. I've tested this with a 2 ohm load, driven with a sine wave to clipping on a scope, and I've driven a 4ohm speaker to full power with the amp bridged. However, be prepared for the following:

- If you use the 5 amp fuses on the rails like he mentions, they'll melt instantly.
- MAKE SURE YOU CAN KEEP IT COOL. The cooling of the amps is designed for an 8ohm load, not a 2 ohm load. To cross 1000W output you'll be dissipating a lot of heat inside the chassis. Active, ducted cooling on the main heatsinks works well, but problem for me keeps happening on the board itself. The transistors degrade rapidly when you don't put anything on them to dissipate heat. In the amp I'm building right now, I have heatsinks on all of the transistors, including little metal tabs that are glued to the small transistors (forget the package style now). Each amp board will have a duct pushing air from a 120mm fan across it to stop me troubles.

 

[Greg B.]

I have a stereo Low Tim II from the late 70's and a quad mono Leach 4.5. They are great amps. Just today I picked up a clean Hafler DH200 at a local electronics swap meet. Ultimately I will do some form of Pooging on the Hafler, but after replacing bad input caps with some polypropalenes and the stock RCA jacks with some decent glod plated ones, I did a shoot out between it and 2 channels of the Leach 4.5. The Leach wins hands down. It is not subtle at all. The Leach has far more definition. Cymbals have that sheen that was completely missing in the Hafler. Massed vocals are are an assembly of individual voices in the Leach. They are a blur with the Hafler. The Leach handles all complex passages much better.

So it's on to modding the Hafler to see if it can beat the Leach 4.5. Hopefully I will be able to determine if it is the topological differences or the details of the execution that differentiate the two.

 

[Damon Hill]

I have a stereo Low Tim II from the late 70's and a quad mono Leach 4.5. They are great amps. Just today I picked up a clean Hafler DH200 at a local electronics swap meet. Ultimately I will do some form of Pooging on the Hafler, but after replacing bad input caps with some polypropalenes and the stock RCA jacks with some decent glod plated ones, I did a shoot out between it and 2 channels of the Leach 4.5. The Leach wins hands down. It is not subtle at all. The Leach has far more definition. Cymbals have that sheen that was completely missing in the Hafler. Massed vocals are are an assembly of individual voices in the Leach. They are a blur with the Hafler. The Leach handles all complex passages much better.

So it's on to modding the Hafler to see if it can beat the Leach 4.5. Hopefully I will be able to determine if it is the topological differences or the details of the execution that differentiate the two.

That was somewhat my experience with my upgrade to 4.5
boards; more depth of field and focus, and individual instruments
and many sounds of all sorts had more 'character' rather
than being, well, generic.

I can still fault massed orchestral works and other complex
music for lacking focus, but I really need better speakers. I
hope to audition my Leach amp on an electrostatic system
some day and see how it does in the definition department.

I am wondering what else could be done to further update
Leach's design, such as Toshiba 2SA1943/2SC5200 output
transistors, current mirrors or whatever. But I'm saving that
sort of tinkering for my 'finished' version with suitable
heatsinks for flatpack output transistors. I don't think much
elaboration on the existing design would be worthwhile; its
relative simplicity may be elegant enough and just some
'voicing' with selected components might be tweak enough.

I'd be interested in hearing your progress with the Hafler.
Two aspects of the Leach design that I've always liked are
the mirror symmetry of the circuitry, and the two stages
of RC filtering in the power supply rails to the input
section.

 

[pooge]

One modification: a 1 uF film cap between the driver
emitters, as per Self and Sloan. Supposed to help
turn off the outputs and improve high frequency
distortion.

Originally posted by Charles Hansen
If you listen you will see that the capacitor here improves the sound quality. It doesn't matter that it's a triple emitter follower. When it comes to making something sound better, thinking is good but listening is better.

Well, I haven't yet listened to this mod. As far as thinking goes, Leach's thinking on this issue has been added to his web site at that bottom of this page:

Leach's output stage

 

[Charles Hansen]

I don't think Leach has listened to it yet, either. Why don't you try listening to it? It will take about 5 minutes. I'm not sure why you want to debate about the reasoning instead of just listening to it.

"When it comes to making something sound better, thinking is good but listening is better."

 

[pooge]

I don't think Leach has listened to it yet, either. Why don't you try listening to it? It will take about 5 minutes. I'm not sure why you want to debate about the reasoning instead of just listening to it.

"When it comes to making something sound better, thinking is good but listening is better."

I haven't had time yet, alright?
In the meantime, why do you have a problem with me trying to learn something, instead of just criticizing me for having the audacitiy to "debate" this issue? If you can hear a difference with capacitor, maybe it's not for the reason given. If there is a difference, I'd like to know why. Can you handle that?
Perhaps it has nothing to do with charge depletion. Maybe it has to do with bias voltage stability or something else.
BTW, it will take a LOT longer than 5 minutes!

 

[Charles Hansen]

Perhaps it has nothing to do with charge depletion. Maybe it has to do with bias voltage stability or something else.

That is exactly my point. Leach is (erroneously) drawing a conclusion that it will make no difference because he performed a "thought experiment" considering only one factor (transistor switching time).

But it does make an audible difference. And that is why listening is more important than thinking. If a circuit change makes an audible difference, then if one wishes, one can spend time to understand why. This is a one-to-many mapping.

But simply thinking about the problem (as Leach has done) can give the wrong answer, as there are too many factors to consider appropriately. He is trying to do a many-to-one mapping, but choosing the wrong starting point and so reaching the wrong conclusion.

 

[fab]

I have a stereo Low Tim II from the late 70's and a quad mono Leach 4.5. They are great amps. Just today I picked up a clean Hafler DH200 at a local electronics swap meet. Ultimately I will do some form of Pooging on the Hafler, but after replacing bad input caps with some polypropalenes and the stock RCA jacks with some decent glod plated ones, I did a shoot out between it and 2 channels of the Leach 4.5. The Leach wins hands down. It is not subtle at all. The Leach has far more definition. Cymbals have that sheen that was completely missing in the Hafler. Massed vocals are are an assembly of individual voices in the Leach. They are a blur with the Hafler. The Leach handles all complex passages much better.

So it's on to modding the Hafler to see if it can beat the Leach 4.5. Hopefully I will be able to determine if it is the topological differences or the details of the execution that differentiate the two.

Greg

Before you quit on modifying the DH-200, please consult this thread
http://www.diyaudio.com/forums/showthread.php?postid=359323#post359323

On my side, I am sad that I have not heard the Leach amp but what I can say is that the mods I have done on the DH-200 amp resulted in having the good characteristics on sound that you mention about the Leach. It is true that the original stock DH-200 sound is clean but not as detailed as the modified one. So, since you want to pooge the DH-200 and you have a Leach, keep us posted on your results.

Fab

 

[pooge]

Leach is (erroneously) drawing a conclusion that it will make no difference because he performed a "thought experiment" considering only one factor (transistor switching time).

But it does make an audible difference. And that is why listening is more important than thinking. If a circuit change makes an audible difference, then if one wishes, one can spend time to understand why. This is a one-to-many mapping.

But simply thinking about the problem (as Leach has done) can give the wrong answer, as there are too many factors to consider appropriately. He is trying to do a many-to-one mapping, but choosing the wrong starting point and so reaching the wrong conclusion.

Leach did not say it would not make any difference. He simply answered that a capacitor in that position will not contribute to charge depletion as alleged by others. If he is in "error" with that, you haven't prooved him wrong simply because you say you hear a difference.

You also stated that Leach hasn't listened to it. How the hell do you know? Is it because if he did, he would just have to agree with you and the modification would have been made? You are really running around the barn to support your end of the "debate". For all we know, he tried it, and it didn't make a difference to him in this amp. Maybe he didn't. Maybe eventually he will. But you sure seem to be making up "facts" as you go along to make your "point".

Leach has made many changes in the amp over the years. If this change makes a difference, perhaps he will change it. But at least we'll know he'll "spend time to understand why".

 

[Charles Hansen]

Let's review. Damon Hill said that he modified his Leach amp by adding a capacitor across the driver bias resistor. You said you didn't think it would help. I said that it makes a clear audible improvement and suggested you try it.

That was good advice, freely given, by someone with years of experience. The advice was based on direct experience in a carefully controlled experiment. Anyone who wants to try it is welcome to. They would have a better sounding amp as a result.

Now you are cursing me in public. It doesn't really make me want to share my hard-earned experience here too much...

 

[Damon Hill]

I'd comment more, but I'm too busy reading up the
various sources to figure out who said what
about which and why or why not.

Is a triple Darlington a triple? Since it doesn't
involve a voltage gain stage, I'm assuming not.

Chill out guys, and heat up those soldering irons
instead.

 

[Charles Hansen]

A triple means there are three sets (usually complementary pairs) of transistors in the output stage. It doesn't matter if they are common emitter or common collector, they still count as a triple.

The "output stage" normally has no voltage gain, only current gain, so that is how you can tell where to start counting transistors for the triple. Most amplifiers are only a double, with two sets of transistors in the output stage. When using bipolar transistors, the extra current gain from a triple significantly reduces the loading on the voltage amplification stage. This in turn lowers the distortion.

 

[pooge]

Let's review. Damon Hill said that he modified his Leach amp by adding a capacitor across the driver bias resistor. You said you didn't think it would help. I said that it makes a clear audible improvement and suggested you try it.

That was good advice, freely given, by someone with years of experience. The advice was based on direct experience in a carefully controlled experiment. Anyone who wants to try it is welcome to. They would have a better sounding amp as a result.

Now you are cursing me in public. It doesn't really make me want to share my hard-earned experience here too much...

I said I didn't think it would help, and then admitted my off-the-cuff error. I investigated further, and obtained contrary reasoning by Leach to refute the reasoning given that the capacitor speeds shutoff. For some unknown reason, you seem to find that insulting or against your philosophy.

The ONLY thing you've contributed up to now is your opinion that the capacitor sounds better (and Damon already implied this). (This in what appeared to be an affirmation of the assertion that it speeds shutoff. ) So how much do we owe you?

When I initiated a contrary opinion from Leach about speeding shutoff, I thought I was making a contribution that could clear up some questions about this capacitor (or create questions to be further investigated), but you started playing superior by cutting down Leach with made up "facts", and criticizing me for not accepting your opnion unquestioningly. You also gave the implication that I'm some sort of degenerate for creating a "debate"! Well excuse me!

You never stated anything about controlled experiments. You didn't even say if your experiments were done on a Leach amp.

And now you say I'm cursing you??? Where? I'm only naturally reacting to your attitude.

If you don't want to share here, that's your choice. If you want to share facts to support your opinions without going out of your way to put down others with a different approach than yours, then your are more than welcome.

 

[Charles Hansen]

How the hell do you know?

And now you say I'm cursing you??? Where?

 

[Jocko Homo]

Typical......

Yeah, just how would you know.

Oh...we are superior in this thread. And holding back your further understanding.

And what was it..........something like "engineers are trained by corporations not to think in unconventional ways".......or some similar nonsense.

(See the zobel network food fight for an accurate quote.)

You guys can't have it both ways when you already have your minds made up.

Pick one excuse to denegrate us, and stick with it.

Jocko

 

[andy_c]

(...)Perhaps it has nothing to do with charge depletion. Maybe it has to do with bias voltage stability or something else.

Have you ever messed around with SPICE simulation, say, with the freeware LTSpice from Linear Technology? By simulating the combination of just the output stage and "class A" drivers, you'll be able to see what's going on. I've done a similar thing, except with bipolar drivers driving a power MOSFET output stage - but the basic idea is the same. I just used some ideal DC voltage sources to set up the DC bias between the bases of Q16 and Q17 (two equal ones in series) and drove the junction of the two with an ideal large-signal square wave voltage source to look at the transient response and switching behavior. Here's what I found.

First, the assumption that the driver is always in class A mode breaks down in large-signal transient analysis with square waves. That's why I put "class A" in quotes above. It looks like Leach's graphs of the currents in the driver were done using swept DC analysis, which will show an optimistic view of the situation. The "gotcha" in this case occurs when attempting to turn one of the output devices OFF. Let's assume we're talking about the NPN output device Q18. It turns out that because of the stored charge in the output device's internal capacitances, under transient conditions there's a brief instant where the base current of the NPN output device is negative (that is, current is coming out of the base of the NPN in order to pull out the stored charge). This current may be tens of milliamps or more, depending on how fast the square wave rise time is. This transient current can be greater than the DC current of Q16 and Q17. So under these conditions, Q16 will actually cut off. With Q16 momentarily OFF, Q17 is now left with the job of pulling the stored charge out of the base of Q18. But it can only do this through the series resistor R36, which slows down the turnoff because of the time constant of this resistor and the device capacitance of Q18. The addition of a capacitor of about 1uF in parallel with R36 fixes this problem. Of course R41 is still having an effect, but it's much smaller than R36.

The reason I even looked at this problem was that when I was simulating large-signal square wave performance of my amp into a capacitive load, I had a horrible overshoot of about 30 percent. I finally isolated the problem to this very issue, and putting a capacitor in parallel with the driver emitter resistor fixed it completely (in simulation).

In short, ths "same sex" driver transistor can momentarily cut off when turning off the corresponding output device. This leaves the job to the "opposite sex" driver, which can do its job better with a capacitor in parallel with the emitter resistor, as this makes it act more like an ideal voltage source.

BTW, I have nothing against Dr. Leach - in fact I was a student of his in the 1977-1980 time frame and built one of his amps.

 

[Eva]

In simulation, I've found that the most effective solution to mantain driver and predriver bias stable during AC transient conditions and to optimize transistor turn-off times [also speeds-up clipping exiting process] is to add RC series networks between bases of predrivers, drivers and output devices

Optimum R and C values appear to be tricky to find, but for example, R in the range 1-10 ohms and C in the range 1-10uF are useful for the last stage [between bases of the output devices]

Placing capacitors without series resistance appears to be less effective since the capacitor tends to charge further on each transient [quickly than it discharges] and this causes cross-conduction and also forces the preceeding stage to go into class-B until the capacitor discharges to the steady-state level. Placing capacitors directly in more than one level of the output stage darlington also appears to increase the likelyhood of parasitistic oscillations during transients [ie : zero current crossing and clipping-exiting] since drivers and predrivers are much 'slower' when they are forced to work in class B or at very low current levels

To see all those effects, output stages should be simulated at frequencies of 10Khz or higher and at high output levels. Simulating clipping behavior is also useful. Below 1Khz or at low output levels these charge-removal effects appear to be almost negligible [the problem arises when faster turn-off is required and base charge-removal current starts to be of a higher magnitude than base drive current and/or driver biasing current]

 

[zinsula]

Eva

Did you the simulations only with bipolars or also with MOS-FET's? Thinking about gate charge (and discharge), with MOS-FET's it would be pretty much the same, what do you think about?

Tino