Hafler DH-200/220 Mods

DH-500

I am back with some report.
Thanks to all your posts I was able to do some changes.

My modifications on the Hafler:

1. To prevent the power switch fails I upgraded it from stock 15A to 20A rated switch.

2. I replaced original RCA jacks with the high quality gold plated RCA jacks.

3. Input wiring was replaced with shielded coaxial cable. That one gave immediate result. Original wiring was quiet only with shorted inputs. Without any cables connected amp produced a noticeable buzz. After modification amp became quiet with open inputs also.

4. Gold plated brass binding posts were installed for the speaker outputs. Bigger holes in the upgraded posts made connection of the 14 GA speaker wire much easier.

5. Power cord upgrade included installation of 20-amp gold plated IEC adapter. This permits usage of commercial or DIY power cords, which have IEC adapters on the component end. I don't believe that different power cords "sounds differently", but if a particular installation permits, I prefer the shortest and thickest (14 GA is a reasonable grade) cord possible.

6. Ground connect/lift switch was added for the ground power wire to minimize possible audio hum in the large audio and video systems like mine caused by a "ground loop". Raising the Ground Lift switch to its up position puts a 10 Ohm resistor in series between the power ground and the chassis. The chassis is still grounded.

7. Bias was set to 350 mA of idle current. DC offset in both channels is less than 10 mV.

A few pictures are posted here:
http://home.comcast.net/~kazarinov/howto/hafler500/index.htm

It is impossible to compare the sound quality of the modified amp with the original (and I am sceptical when somebody says that he remembers how amp sounded about one week ago), but it sounds better than my Rotel RB985 now.
 
mishak,

Congrats!! Nice work on the upgrade of your DH-500.

If you compare the schematic/parts list of the DH-220 to that of your DH-500 you will see that the circuit cards are identical, save the values of two resistors on either side of the input DC null circuit.

The DH-200 was the introductory offer by Hafler to the DIY market and it was built to some low price points and scimped on the number and quality of certain parts, mostly film bypass caps. The DH-220 incorporated most of the mods published about the DH-200, to the point there is little one can do, simply, to make hearable improvements in the DH-220.

The same now holds for the DH-500. You have upgraded the last weak links by installing better quality input and output connectors.

The DH-500 includes an MOV in series with the incoming AC which is supposed to protect the ON/OFF switch from initial current surges at turn on to the point the stock switch supposedly can survive most conditions. Your 20 amp switch should provide great peace of mind.

Now, if you want to further improve the sound I can testify that the Musical Concepts PA-3C circuit cards make a big difference. I have carefully compared the sound of a stock DH-220 to a similar chassis that has the MC circuit cards and pair of 26K uF filter caps installed. Hands down, for my music tastes, the MC cards are superior. I would imagine the same sonic differences would be present for the DH-500. But, these mods are not cheap.
 
Dick West,
Thanks for your comment.

The major upgrade using the Musical Concepts boards for the DH-500 is somehow not obvious for me.

First, that would replace the part that I am concern less and keep parts I dislike more (ugly box, loud fan, inefficient internal space usage, etc.)
Second, performance/cost factor. Untouched original DH-500 costs $250 - $300, golden connectors another $50, Musical Concepts boards $350. Total $700 plus work and time.

The same age Bryston 4B on Ebay costs $600 - $800 with far superior (as they all say, or it is not true?) sound quality. So I am going to wait till my local Hi Fi audio shop would get used Bryston, take it home for the A/B comparison with DH-500, and only after that make my decision what to do with Hafler.
 
Administrator
Joined 2004
Paid Member
Hi mishak,
As soon as you hear the Bryston, you will understand why people buy Haflers. The early Bryston 4B sounded awful and has technical defects on top of that. People say a lot of things that are untrue. BTW, don't leave the Bryston on unattended. There is no speaker protection at all and I have seen them go DC all by themselves. They will take about 1/2 an hour playing music at a good volume before they start to sound like an amplifier. They get very hot, so provide at least 6" clear above.

Keep in mind that some clear air space inside the chassis is needed to cool other parts. It isn't wasted.

Also, don't forget that when you buy an older piece of equipment, it will need work. None of these will sound as they did when they were new. Dick has given you some very good advice, and he has nothing to gain from this.

-Chris
 
anatech said:
Hi mishak,
As soon as you hear the Bryston, you will understand why people buy Haflers. The early Bryston 4B sounded awful and has technical defects on top of that. People say a lot of things that are untrue. BTW, don't leave the Bryston on unattended. There is no speaker protection at all and I have seen them go DC all by themselves. They will take about 1/2 an hour playing music at a good volume before they start to sound like an amplifier. They get very hot, so provide at least 6" clear above.
-Chris

I understend, so why I am not selling my Hafier, and I want to put it back to back with 4B, maybe 4B NRB, and see, which one suit me better.
If Bryston is not obviously better, that would be a good point to consider the next modification.
 
Administrator
Joined 2004
Paid Member
Hi mishak,
I just wanted to explain why I am not rushing to execute all of them immediately.
That's fair enough, and wise in my book.

All of us older folks have long experience and you should be free to learn some things on your own.

From what I've heard, the Bryston SST versions are the only ones that sound any good. I haven't heard one myself, but I've listened to more than my share of 4B's. I have serviced both the Bryston and the Hafler amps.

-Chris
 
Mishak dh500 mods

Hi mishak,i have been watching your progress with these mods,and just wanted
to say well done, but there is much more that can be considered yet!
I have been systematic in my writing so as to break down the many improvements
possible without too much confusion to those less technically adept, its sometimes hard to get the ideas in words, you always seem to miss something. I will
summarise all my suggestions at the end of my recital..
Just as a matter of interest the old Hafler 200/220 and 500 were always prefered
sonically to the later 280/600 series even the distributers here thought the same
despite the so called advanced linear advancement of the predrivers etc.
I will be dealing with a simpler method of doing this later.It would appear that there
are a large number of{technical audiophiles}who do not like FETs in the input
circuit because it is claimed they add a graininess to the sound of the high frequencies despite measurement contradictions, the later haflers went over to
FETs in the I/P stages. regarding your 500 do make sure the fan is in good order
that heatsink assembly does generate a fair heat;finally do watch my comment on the way you set the current using the meter in series, to my cost i have caused
instability with this method on all sorts of amps the length of the leads act like
massive aerials and MOSFETS can get very twitchy, its no fun having to select
matched transistors throughout a hafler board because they have been destroyed
by spurious oscillations. My best wishes to you
regards
 
Administrator
Joined 2004
Paid Member
Hi humble,
It would appear that there are a large number of{technical audiophiles}who do not like FETs in the input circuit because it is claimed they add a graininess to the sound of the high frequencies despite measurement contradictions
Well, you just found one (technician) that will say the opposite is true. J fet inputs normally lend and "openness" to the sound. I find that it's normally the BJT inputs that sound grainy. The complimentary diff BJT input seems to sound very grainy.

Give me a J fet diff pair any day over a BJT diff pair. ;)

-Chris
 
Re FET I/P stages

Thank you anatech for your input,I to like to debate for the fun of it,however
I always respect peoples preferences,If you like thermionic triodes and single-ended
o/p that is fine with me I am only reporting the opinion of many of my colleages
who have over many years hunted for the holy grail in sound.technically speaking
most FET I/P circuits tend to be rather more complicated to get the margins of
linearity lower than for a much simpler BJT which I am sure you are aware. however
direct comparisons are available from different sources, one such is in the Hitachi
application notes in which an all FET 100 watt amplifier was offered as an alternative to a BJT equivelent both used mosfet o/p.The all FET design used twice as many
devices to achieve the level of distortion up to about 20khz then above that the all FET design started to increase its distortion so that at 50khz it was five times
worse;as I stated many people regarded this sort of behavior as graininess,some say
hardness and other expressions of dislike.However we all know, dont we that
FET's are square law devices and BJT just exponential,the corollary being that all things being equal the BJT is more linear and will require less feedback etc,
I naturally defer to your preference for FET I/P stages, but technically I am
forced to disagree
regards
 
Administrator
Joined 2004
Paid Member
Hi humble,
:)
:D
:rofl:
Sorry. I guess there is a misconception there. A big one actually.

A square law device is more linear than an exponential device. A differential fet pair (or tube) has a far more linear range than BJT's. If you degenerate the emitters in a BJT pair to reduce the transconductance, you can get closer to a more linear transfer characteristic - like fets and tubes. 'Tis the truth of the situation - technically speaking.

The cost of degenerating the emitters of a BJT is a loss of gain and higher distortion. Neat eh? Sounds like fets or tubes. However, the sound quality is often better.

Now, the sound of an entire amplifier has much to do with the topology as well as the input pair. A J-fet diff pair can be made exactly the same as a simple BJT setup. It will sound smoother, all other things considered equal. One of the errors you made was to compare one extreme example of a circuit with another. No averages, just one specific case. Also, keep in mind that raw distortion numbers are not a great indicator of sound quality taken alone.

I'm not attempting to debate you on this. Over time you will discover the truth - the physics of the situation. In the mean time, enjoy the journey. There are also a number of threads that address these issues if you do some searching. This site is a gold mine of great technical information.

My preferences include tube, fet and some BJT input amps. All I care about is that they sound good. I do own all three types and will switch between them. They all bring something different to the table to enjoy.

-Chris ;)
 
i'll apologize in advance - i don't have the Hitachi app notes in front of me right at the moment, but I've seen them and built those amps YEARs ago ...

i agree with anatech ...

i'm not sure those circuits (the all-fet driver versus the all-bipolar driver) were designed with similar target requirements (open and close loop bandwidth, distortion, etc.), so trying to do an apples to apples comparsion would be misleading i think. At the very least, I am thinking the drivers to the output MOSFETs were not even running at the same current, which could be a contributor to differences at the high end right away.

Better approach for comparison is to:
1. establish the performance requirements
2. have the designers chose their weapons: fets or bipolars
3. have the designers submit their "optimized design" amps (by optimized, i mean take advantage of the chosen technology and chosen topology strengths and compensate/mitigate for the weakness)
4. have some wine and have the listening shootout
;)
5. submit findings to diyaudio and enjoy the resulting show
:smash: :D

There was a regulator design showdown in "the other place" that kinda, sorta followed this process. worked out to be pretty interesting.

BTW, if we're voting, among optimized designs with bipolar front ends and jfet front ends, I prefer the jfet myself. I like the XL-280 better than the DH200 or DH220. I've been curious about the XL-600 schematic to see if it's the same as XL-280 but I've never been able to see it. Curiously, the XL-600 manual on the Hafler site doesn't include the schematic.

... All I care about is that they sound good. I do own all three types and will switch between them. They all bring something different to the table to enjoy.

-Chris


amen, chris!
mlloyd1
 
FET's versus BJT's I/P stages

Hi Anatech, thank you for your response,I guess we went to different tech
collages for my experience in this matter differs radically to your own it would seem,
I think the best thing i can do with this question is to quote directly from a well
known designer called Randy Slone from his book High-Power Audio Amplifer
Construction Manual[isbn0-07-134119-6] page 90 refers..To begin JFETS
cannot perform as well as BJTS for the function of a differential input stage.To state
it bluntly,their transconductance is lower, THEIR LINEARITY IS POORER, and their
Vgs parameters vary by a much broader range when compared to even mediocre
BJTS... the only real advantage offered by a JFET input stage is the elimination of
any DC offset errors at the input. end of quote, he also does measurements to
illustrate these facts, my own experience in audio design mirrors these findings
which with respect puts the matter beyond debate as far as i am concerned, I therefore stand by my previous comments.
regards
 
Administrator
Joined 2004
Paid Member
Hey humble,
No problem. I'm not trying to convert you here.

Just a couple things to remember. A device used improperly will always perform poorly. Quoting from a single author is much like walking with blinders on.

I would like to call your attention to one error you did make earlier on.
FET's are square law devices and BJT just exponential
The transfer characteristic you are referring to and conclusions you make are odd. If you put both a square law and exponential on the same graph, you will see very quickly that the exponential device is less linear than a square law device.

To begin JFETS
cannot perform as well as BJTS for the function of a differential input stage.To state
it bluntly,their transconductance is lower, THEIR LINEARITY IS POORER, and their Vgs parameters vary by a much broader range when compared to even mediocre BJTS
Often, you will see a BJT stage with the emitters degenerated (to lower their transconductance) so they are more linear. The VGS parameters don't matter in a matched pair ('cause they are matched) and finally, we do tend to run the tail current from a CCS.
the only real advantage offered by a JFET input stage is the elimination of
You just contradicted yourself (or someone did). Lower offsets mean better error correction. DC offsets can be lowered by matching the transistors (either type) and lowering the base current of the BJT, or injecting a compensating current to return the DC offset to zero. Many 70's amplifiers were designed this way.

I find that it is very important to read from many sources as you can not expect every author to be correct on all counts. This actually seems like Mr. Sloane was reaching to justify a conclusion based on his own prejudices rather than a balanced technical argument.

I design with both fets and BJT's for differential input pairs. Which one you use depends on many things. Sometimes it's just what I feel like using. BJT's are almost always degenerated when I use them. Check out the SymAsym threads for some viewpoints from both camps.

-Chris
 
FET verses BJT postscript

HI again Anatech, the term exponential that I use is to generally describe the characteristic slopes for say a triode or a BJT to distinguish from the more severe
[rounded slope ,square law] of say a pentode or a FET; FETS are often compared to
the characteristic slopes of the pentode, the pentode is often run as a triode
because the slope is straighter and therefore gives less distortion.I had intimated
in my earlier comments that in general if you use FETS in the I/P stage it is
almost certain to be more complicated to achieve the levels of distortion that can
easily be achieved by a much simpler BJT setup, so to points..If you use a pair of
fets fed with a ccs and a current mirror what the spectral analysis will show for this
circuit is a distribution of harmonics in which many of the odd harmonics created
are higher in amplitude than their even counterparts with a definite tendency for
rather larger distortion figures in the higher frequency domain something like 0.2
percent around the 20khz region, by contrast an equivelent BJT setup will have
an even decay in distortion harmonically and will demonstrate at 20khz a distortion
figure some 20 to 50 times lower, yes you can make FET circuits perform well
as you intimate with care and more components, the point is why bother when
a simpler circuit does the job, remember that many comercial designs are made
down to a price with smallest component count for the most profit and usually
only measured at 1khz for its distortion figure of merit, the purist design concepts
are a different matter altogether.I had simply stated that many of my colleages
did not care for those FET designs from the commercial stables and i agreed
the difference being that I understood why,what to look for and how to measure
it, and that is not at one frequency, but many spectrally.
regards
 
Administrator
Joined 2004
Paid Member
Hi humble,
Ahh, I was referring to the tranfser characteristic of the raw device. The exponential curve of a BJT being far less linear that a fet or tube.

The measured THD of an amplifier will be higher for a fet than a BJT simply due to reduced transconductance. The harmonics will be similar, but higher for the fet for the same reason. However, the BJT will clip far more easily unless emitter degeneration resistors are used. You also need to attenuate the input signal first. The changing input impedance of a BJT may predistort a signal if the source impedance is too high. So the signal has been damaged before it even enters the amplifier. This is why both Marantz and Nakamichi (possibly others) came out with buffers between the diff pair and the outside world. Marantz did this in 1968 when designing the 500, Nakamichi in the 90's. Same principle.

At any rate, some further reading may be in order. Your experimental results are at odds with mine (and many others) and it may simply be the conditions the experiments were made under. If you have a preconceived notion that fets are worse than BJT's, you may set things up to prove you are correct. I do my best to keep an open mind and compensate for any issues with the devices I use.

At any rate, each device type has a set of advantages and disadvantages. Used properly, each can perform well. Used improperly, none of them will deliver the performance they are capable of.

-Chris
 
Current debate

Hello Anatech,after many words what did we achieve? I am retired now but over many years I debated in WW, now Electonic World, and many other periodicals,I have
discussed all sorts of issues with the best engineers here from polarized cables to fuses, from power supplies to Jitter in CD players, you name it.The only thing I learned was that everyone has his favourite hobby horse, or defers to some particular philosophy or person and it will make no difference what you say, you will change
nothing, i have brought people to my home to demonstrate some aspect of audio
design as an improvement, or some simple mod that can be easily introduced only
to have these ideas put down in some way, some of these ideas have been given
to manufacturers, one i recall was to a famous company beginning with Q regarding
their power supplies actually on both models of their amps but in a different aspect,
For my trouble i received a lecture being told that the amplifiers fulfilled the prime
function of their testing criterior, and that only double blind listening tests should
be used in any evaluation, since the amps fulfilled the test criterior, no such mods
could make any difference, that anyone who didn't believe in double blind listening
tests had feet of clay and had their head in the clouds..Later those mods were
incorporated.Anyway I don't generally debate any longer, people nearly always say
"I thought you meant something else" or widen the goalposts, change the critereor,
We all need to be justified in some way, I have over too many years gathered a large
experience in audio, have wanted, to share with others all the anechdotes and technical gen, but it is futile. history repeats itself and with each generation all the
old arguement reappear,I have been there before,long before now we have the old
favourites of valves resurfacing and BVD'S and turntables,I am glad to be able to
help with some of these matters from my young friends,so now my friend I leave you with the floor and wish you luck
regards
 
Administrator
Joined 2004
Paid Member
Hi humble,
Okay. ?

I was really hoping you would have actually explained your position more clearly. I am ready to consider any technical data that you may have.

I have no doubt that you have satisfied yourself in your beliefs. You are correct in that I am a youngster though!

Cheers, Chris
 
Last words

Well Anatech,you may be young but you represent our future,All power to you.
continue to search for your truths, which is a truly hard journey,the best advice
I can leave you with is to paraphrase my old mate BEN DUNCAN..Always check
your references,and stay focussed.
Kind regards