Hitachi Amp Fusible Resistors (and general servicing)

[Jez1234]

A Zobel network is so often confused with the damped series inductor of a few uH at the output of most solid state amps. Naim amps have a Zobel network but not the resistively damped inductor (which it has become "trendy" to call a Thiele network....).

Negative feedback audio amps (and linear PSU regs etc) look inductive at their outputs and this can interact with any capacitance forming part of the load to create an output pole at the feedback takeoff point, which can and often does then cause oscillation. The highly damped inductor offers a degree of isolation at HF between amp and capacitive load.
Interestingly, amplifier reviews sometimes comment on square wave oscillographs showing overshoot and ringing into capacitive loads which are actually caused by insufficient damping of the output tuned circuit so formed!

The lack of output Thiele network in the Naim (and some other models such as Exposure... which basically ARE Naim ) strikes me as the original models "tried to be clever" in eschewing it but upon it being quickly found that a capacitive load could cause instability it was grasped onto as a marketing feature in that it was "designed" to work best with certain speaker cables and that rival models were ultimately handicapped because they do have a Thiele network Unexpectedly the recommended cables (originally an RS speaker cable IIRC) were to be of a certain minimum length... which just happened to provide sufficient inductance to mimic the effect of the missing Thiele network.... When they later marketed their own brand expensive speaker cable then the sales ploy was completed....

[Trevor]

Jez an exposure is not a Naim it uses inverted darlingtons in the output stage thus isolating the Darlington element from the bias network It does use a diff pair at the input a
simple vas stage and phase compensation i knew John Farlow well and sold his products
John was far cleverer than I thought and he even sold his company on so that he could retire with money in the bank ! something that very few audio engineers ever managed
have personally made and sold on many amplifiers that do not use the inductor resistor combination network as they would of course show ringing etc under test according to the scope you are looking at the combination not the amp
Audio is in some ways so simple but also so complicated
Trev

[knobtwiddler]

Originally Posted by Radio Wrangler

Those funny earthing schemes like single-point are OK for avoiding hum problems, which is their reason for existing, but they can go seriously astray at higher frequencies and can contribute to instabilities.

I still don't get why Hitachi would common the 0V at the front end. The only justification in my mind is due to the headphone O/P. If they had separate PSUs for L/R, why didn't they keep each channel wholly independent? It seems like a recipe for a ground loop to me. The Hitachi keeps signal 0V separate from chassis ground.

That's an interesting development of the DH200. Cascoding the differential input stage is good, but he fails to do the same to the VAS. I went a bit further and had differential, cascoded VASs in the positive side and the negative side rather than current mirrors.

Cascoding reduces phase shift, extends frequency response and reduces the Early effect. It also takes the high voltage off of the transistors Vce and you can use separate high gain and high voltage parts. I was also trying to use as many transistors as possible (subject to them doing something useful) as a joke with a friend.

David

Out of curiosity, does your amp have Lateral MOSFET OPS?

Thanks again for the wisdom!

Exposure - that's a blast from the past! I always liked their stuff. Had a dealer on my doorstep. I don't remember too much about what went on inside it.

[Radio Wrangler]

I did three crazy amps, one was valved and output-transformerless. The other two were closely related, one with Hitachi lateral MOSFETs, the other with bipolars. In the end I chose to keep the Mosfet amp and boiled the others down to re-use the bits. A small number of them were made at HP. The amp Bill designed was somewhat minimalist in comparison and wound up being made at Linn after he moved there. Great fun.

David

[G.Castle]

Re. Post 22... what is an 'inverted darlington'? Never heard the expression before.

Thanks.

[MelJon66]

I think it's marketing speak for a Sziklai pair which uses one PNP and one NPN in a Darlington type arrangement. Darlingtons use two of the same type and are often in one package. Sziklai pairs always seem to use discrete transistors.

If the more knowledgeable members would care to add to that (or correct it) I would be pleased to hear it too.

Cheers
Mel

[Trevor]

No its not marketing speak in most amplifiers that use darlingtons they are used as emitter followers this is ok but the bias stability is poor if you invert them and put the load in the collectors the bias stability is improved but it does cost you a few more transistors
If you look up Linley hoods 30 watt amplifier you will see its implementation in a schematic form

[G.Castle]

Thanks both.

I did wonder if it referred to Sziklai pair myself. So nothing special, just darlingtons.

[knobtwiddler]

Originally Posted by Radio Wrangler

I did three crazy amps, one was valved and output-transformerless. The other two were closely related, one with Hitachi lateral MOSFETs, the other with bipolars. In the end I chose to keep the Mosfet amp and boiled the others down to re-use the bits. A small number of them were made at HP. The amp Bill designed was somewhat minimalist in comparison and wound up being made at Linn after he moved there. Great fun.

David

It's fascinating that you kept the Lat-FET amp. Unlike their BJT-OPS brethren, where plenty of techniques can be employed to give good THD performance while maintaining sane levels of bias, my understanding of Lat-FETs is that there is little in the way of trickery that can be employed to prevent x-over distortion. The best you can do is find a sweet spot so that it keeps Class A up to a moderate level, and you get a smooth transition into the Class B. But this is always going to involve a fair amount of standing current, so the amp will need sizeable heatsinks and will be sucking a lot of pwr.

I've trimmed amps depending on speakers used. I've found that the Hitachi-suggested figure of 200mA is about the bare minimum you want. I believe the Cordell DH-220C uses 200mA per-device. I have an amp in front of me that has +/-70v and was intended originally for studio monitors, with 650mA bias (I suspect I could find a sweet spot at lower bias now, as I have better test gear).

While there have been numerous attempts to improve the VAS stage in Lat-FET amps over the years (the Cordell being the most recently published amp to my knowledge), the P-P OPS stage hasn't seen much change, relying on brute bias. Whether you have a rudimentary, obligate-class A VAS stage such as the 7500 here, or a much more elaborate setup such as the one in the Hitachi HMA-8500 MK2, the dominant THD generator by far will be the OPS. So, even if your VAS has a grillionth of a percent THD, the likes of Douglas Self will tut at your choice of Lat-FETs. I am quite curious as to why a designer with such an objective grounding would choose a Lat-FET amp? I feel a little guilty about mine. But I like the simplicity. There are BJT-OPS amps that have comparable THD at far lower bias, but the component count vs a Lat-FET amp goes through the roof. The Lat-FET (to my knowledge) is the only 'class' of semiconductor ever developed solely for AF amps. My most recent from-scratch quad were built in 2000 and have roasted at 65-degrees, only needing switch-cleaner on the bias switch. Alas, they are massive, ugly and the toroidals have become mechanically noisy. I hope to use the 7500 until I can solve these issues.

[Radio Wrangler]

This amplifier could never be accused of simplicity. At 1dB below clipping I was getting all harmonic products more than 100dB down on the fundamental. Dropping the level saw the harmonics fall faster than the reduction in the fundamental - at the expected ratios. Similarly with two equal tones, peaking 1dB below clipping, odd order intermod products were more than 100dB down on the peak envelope power, and reduced as the level was reduced. I had to make some special test gear to find these results, and when reducing level, I used a correlator to follow them a bit below the noise floor before I lost sight of them. The amp has very wide loop bandwidth and relies on feedback to reduce crossover distortion across the audio range. Distortion creeps up, but only above audible frequencies. Done just for the hell of it for myself, I have had no interest in commercialising it. There are too many corners just waiting for a bean counter to cut and the MOSFETS are no longer available and 40-odd years ago the audio market was going seriously insane (and still hasn't recovered) so I had no intent to leap into that bear pit. It's good enough for me and I've spent 40+ years listening to music, riding horses, building fast things, and getting on with RF work.

The non-linearities of those MOSFETs are no worse than valves, their distortion tends to be low order and handling it with feedback also handles variations in it with temperature. The Hitachi-derived designs were almost all based on Hitachi's application notes where Hitachi was showcasing how amplifier circuits could be made simpler and cheaper with their devices. They were a marketing/sales tool, but not a bad one.

David

[Trevor]

I can remember the launch of the Hitachi units and our shop ordered one I have always liked them ! Sold loads of them in the end I must have owned my 7500 for 40 years or so

[Whaam68]

My humble Hitachi HA250 proudly boats "Inverted Darlington Circuit" written in very big letters on the front panel.....not something that I would have thought would have been a big draw for most prospective customers! That said it's a very nice sounding (non-mosfet) amp.

[Trevor]

Ah yes We sold a lot of them as well ! As the salesman and engineer I would tell my potential customers that this was thermally more stable and should give a good long life

[Jez1234]

Originally Posted by Trevor

Jez an exposure is not a Naim it uses inverted darlingtons in the output stage thus isolating the Darlington element from the bias network It does use a diff pair at the input a
simple vas stage and phase compensation i knew John Farlow well and sold his products
John was far cleverer than I thought and he even sold his company on so that he could retire with money in the bank ! something that very few audio engineers ever managed
have personally made and sold on many amplifiers that do not use the inductor resistor combination network as they would of course show ringing etc under test according to the scope you are looking at the combination not the amp
Audio is in some ways so simple but also so complicated
Trev

Sorry but you are wrong. The early Exposure amps are an EXACT copy of the Naim NAP250 amp design even down to EVERY resistor and compensation cap etc etc being identical and all the small signal transistors being identical Ferranti/Zetex/Diodes Inc/whatever they are calling themselves this week. The complementary output stage "bolted on" in place of Naim's quasi complementary stage is the only difference. I have personally reverse engineered both amps so can guarantee this to be true

[Trevor]

I know that the front end is very similar but the Drivers were very low powered devices with Darlingtons as the outputs There was an IC also that was used as dc fault detection
I knew John in 1979-1980 era and even called into his Hove workshop It was very small scale at that time

[Whaam68]

You’d e been right Trev it’s only just had its first re-fresh 45 odd years later!

[Jez1234]

The Exposure amps use (it varied over the years but for the later ones) a Sziklai pair (aka CFP, Complementary Feedback Pair) made up of Semelab transistors (not Darlingtons). The drivers are SM2177A and SM2178A (4A, 30W, 200V, 100MHz TO220) and the output transistors are a bit special... marked as "EXPP" and "EXPN" they are also Semelab and although I know what they are I believe Exposure consider the information proprietary. They have impressive specs and IF you can find them are very expensive... as much due to their rarity as to their specs...like £50+ each type expensive!
I'd feel fairly confident in guessing that they bought in loads of them whilst they were (pun alert!?) a current device and only paid a couple of quid each for them... Earlier versions used more prosaic output devices such as MJ15003/4 at one time IIRC.

FWIW the "Regulated" models are not, as such regulated. The pedant in me must point out that they are stabilised rather than regulated. A brute force "amplified Zener" arrangement is used, employing the same Sziklai pair as used on the amp boards but with TIP31C/32C as current limiters (current monitored by V drop across small value series resistors which turns on TIP's which rob base current from Sziklais. Pretty standard). DC max current is only around 1.8A but they allow far greater current for short periods such as musical peaks etc.
In the TOTR monoblocks I last worked on the Voltage overhead on the stabilisers was way OTT! About 90V in for a stabilised 40V out Equally bizarre was the use of six 35A bridge rectifiers in parallel per channel! One is fine on its own and the chances of anything near actual current sharing happening in practice strikes me as doubtful.
I vaguely recall seeing some models with a relay and prob a DC detect circuit in some model years (IIUIC there has been various board variants around the same basic circuitry in some models, some incorporating DC protection) but many/most had no such thing fitted and in fact the last ones I repaired had "gone DC" and taken out a woofer...

As I said, everything up to the OPS on the amp modules is not just "very similar" to but in fact an exact copy of the Naim NAP250 down to resistor values, capacitor values, small signal transistor type, Zeners etc etc. Unless my comments on this elsewhere around a year ago have started a Chinese whisper you'll not read this anywhere else as it appears to have been be a well buried "secret"

All comments refer specifically to Farlowe era models such as the 4, 4 Dual Regulated, various TOTR monoblocks (quite rare) and the little unregulated 8.

[Trevor]

As I say I only refer to the 1979-1980 period when \i sold his products He later went on to do the regulated series I never purchased them so I have no knowledge in that case
The information you have provided is very gratefully received ! I can remember being in his workshop and seeing the schematic diagram and noted the supposed phase correction on the drivers having said that I had not seen the Naim designs at that time
As ever we are always learning
Trev

[knobtwiddler]

For interest's sake, I've attached the diagrams for the later Hitachi HMA-8500 MK2. It appears to have slightly less THD and 25 more Watts. It's definitely not a simple design like the 7500. I don't imagine it's any quieter than the 7500, and it appears to have quite a bit of unobtainium silicon. Its weight is about the same as the 7500, so even though it has 2 heatsinks, I suspect they are more lightweight (the 7500 mk2 weighs almost 2 Kg less than MK1, but it also has 2 heatsinks rather than one).

The I/P stage is a very different proposition on the 8500. But is that where it gets its slight THD advantage from? In reality, I wonder how much better the 8500 would be?

edit - I give up. I can't figure out how to compress the 8500 doc down enough to attach. Anyone curious enough will have to get it from Hifi Engine. Sorry!