John,
Your efforts are very much appreciated and I am sure that enough people are paying attention to what you say/point to. Unfortunately, the real problems are much much deeper.
Audio is not hip anymore. People have other toys these days.
General public is too brainwashed - they believe in what marketing tells them.
The level of knowledge is on decline. Ignorance is bliss these days.
The gap between bad audio and good audio is getting wider every day.
Good sounding audio is too expensive to be considered for purchase by "normal" people.
That "certain percentage of people who want it better" that you mention is probably very very small. It is already served by you and a couple of others.
"Good" parts that could be arranged in "alternative topologies" and considered for mass manufacturing are dissapearing at alarming rate.
RoHs, SMD, manufacturing abroad and so on ... Depressing, isn't it?
Do I sound like an old fart?
Your efforts are very much appreciated and I am sure that enough people are paying attention to what you say/point to. Unfortunately, the real problems are much much deeper.
Audio is not hip anymore. People have other toys these days.
General public is too brainwashed - they believe in what marketing tells them.
The level of knowledge is on decline. Ignorance is bliss these days.
The gap between bad audio and good audio is getting wider every day.
Good sounding audio is too expensive to be considered for purchase by "normal" people.
That "certain percentage of people who want it better" that you mention is probably very very small. It is already served by you and a couple of others.
"Good" parts that could be arranged in "alternative topologies" and considered for mass manufacturing are dissapearing at alarming rate.
RoHs, SMD, manufacturing abroad and so on ... Depressing, isn't it?
Do I sound like an old fart?
Electroj, of course you make a good point, but all is not lost. Of course, most people are not going to buy something that costs many thousands of dollars, when they can buy something mass produced, that is 'good enough'. However, here, on this website, you would think that suggestions on how to do something 'better' would be appreciated. Many here are reasonably well educated, show SOME interest in audio. Why should they waste their time making something that they could easily buy off the internet?
What I try to convey, when I get a chance, is what can be done by individuals to make an audio product, better than the typical.
What I try to convey, when I get a chance, is what can be done by individuals to make an audio product, better than the typical.
The HCA 3500 is a design that measures fairly well, yet FAILED in the marketplace. As you can see, NOTHING has been skimped, this unit has lots of parts, features, and a relatively low price. Let's go through this design and try to make it SUCCESSFUL to audiophiles to such an extent that it can 'beat' just about everything.
The price and the story. Double or triple the price and make sure that there are obvious "name" components. Make up a story about how the amplifier eliminates the umbrella effect that greatly inhibits natural sound reproduction in "midfi" amplifiers; maybe throw in some stuff about superior performance by the Geddes metric. See if you can make the protection circuit discrete so that the advertising copywriter or dermatologist who's doing the magazine review doesn't see any opamps. Ditto the servo. Bias the ground-plane of the circuit board- there's scads of story-telling you can do about that.
If you want to be truly cynical, see if you can make the output impedance higher and a bit more nonlinear so that the performance is highly load-dependent. That's part of the entertainment value.
I'm right behind you on this ....
Do you believe it failed because of cost vs Brand recognition ? I have friends who swear by their HCA3500 and this is going back 12-15 yrs.
I'll skip the cynicism...
The impedances at the inverting and non-inverting inputs are not equal which could give rise to what some have called common mode distortion.
The rectifier diodes should be soft recovery types to minimize the switching pulses at several hundred KHz. Can't tell if there is a damping network across the transformer secondaries or not.
The power supply filter caps have many and varied bypass caps connected across them. Inevitably resonances are produced with this sort of arrangement, which leads to higher impedances at some frequencies for the power rail. Better to use a single good cap or to change the bypass arrangement to a low pass filter type. Or, at least add some form of damping of the right value.
There is only a single rectifier bridge used for the two rail supplies. This can cause problems, especially with high efficiency toroidal transformers.
Zener based regulation is noisy. Bypassing a Zener diode with some caps in parallel gives a bumpy noise output over the frequency range; not always helpful. Single pole filter solutions aren't really a great answer in most cases.
Can't tell about the high voltage rails...
Is the AD711 linear enough? Since there is no additional servo filtering after the AD711, any junk produced by the AD711 will just be passed through to the amp's inverting input. It's attenuated by the feedback divider, but still is there. I don't know enough about how well the AD711 performs in this application to really determine that.
Of course, from the schematic there's really no telling about many of the component types, the layout, or the wiring scheme used. But, I bet it simulates just great with the overly simplified passive component models most folks use.
The impedances at the inverting and non-inverting inputs are not equal which could give rise to what some have called common mode distortion.
The rectifier diodes should be soft recovery types to minimize the switching pulses at several hundred KHz. Can't tell if there is a damping network across the transformer secondaries or not.
The power supply filter caps have many and varied bypass caps connected across them. Inevitably resonances are produced with this sort of arrangement, which leads to higher impedances at some frequencies for the power rail. Better to use a single good cap or to change the bypass arrangement to a low pass filter type. Or, at least add some form of damping of the right value.
There is only a single rectifier bridge used for the two rail supplies. This can cause problems, especially with high efficiency toroidal transformers.
Zener based regulation is noisy. Bypassing a Zener diode with some caps in parallel gives a bumpy noise output over the frequency range; not always helpful. Single pole filter solutions aren't really a great answer in most cases.
Can't tell about the high voltage rails...
Is the AD711 linear enough? Since there is no additional servo filtering after the AD711, any junk produced by the AD711 will just be passed through to the amp's inverting input. It's attenuated by the feedback divider, but still is there. I don't know enough about how well the AD711 performs in this application to really determine that.
Of course, from the schematic there's really no telling about many of the component types, the layout, or the wiring scheme used. But, I bet it simulates just great with the overly simplified passive component models most folks use.
Well let us start with the stuff you still don't get right!
There are three basic power supplies. All of them use center tapped transformer windings where the center tap is grounded. Now it doesn't show the power transformer but if there are three separate windings then each center tap will have a different amount of coupling to the primary winding and these current differences will become differential currents on the ground conductors.
Now the noise that couples is some power line fundamental frequencies, but more annoying is the higher frequency noise from dirty loads on the rest of the AC power system.
Even with a single tapped secondary that noise will still make it to the audio reference ground and have a secondary path to the preamp through the input shields in most cases.
Now on the power supply to the input stage. It has a 600 volt bridge on what is listed as a 75 Volt AC supply. Allowing for 20% high line and 10% transformer regulation that would place 280 volts across the bridge so using a 600 volt rating instead of a 400 volt unit will increase the diode turn off noise.
Then the bypass capacitors are ganged in parallel too close together in values, not to mention the series use of the 4700/63 main filters. As these are typically 20% parts during discharge the voltage across the capacitors almost certainly will not be equal. This will result in different discharge currents and additional non-linear reactance of the power supply rails.
Then there is the electronic regulator which at least uses resistors to try and stabilize the capacitance balance issues, but then also does this to the bypass (some of which are excess) caps and that only serves to raise their inductance and reduce their effectiveness.
The zener diode is directly in parallel with the series regulator filter by using an isolation resistor or inductor you get more effective filtering.
Now the other real issue as I read it is that SW1 changes the overall gain of the amplifier. That path goes through connectors and who knows what in actual implementation. The switch and all the connectors add a small bit of contact distortion.
Later we can talk about the rest of the inadequacies.
But the power supply issues can be huge because you bench test with a clean line and the user rarely has one. So you can measure well and deliver ....
There are three basic power supplies. All of them use center tapped transformer windings where the center tap is grounded. Now it doesn't show the power transformer but if there are three separate windings then each center tap will have a different amount of coupling to the primary winding and these current differences will become differential currents on the ground conductors.
Now the noise that couples is some power line fundamental frequencies, but more annoying is the higher frequency noise from dirty loads on the rest of the AC power system.
Even with a single tapped secondary that noise will still make it to the audio reference ground and have a secondary path to the preamp through the input shields in most cases.
Now on the power supply to the input stage. It has a 600 volt bridge on what is listed as a 75 Volt AC supply. Allowing for 20% high line and 10% transformer regulation that would place 280 volts across the bridge so using a 600 volt rating instead of a 400 volt unit will increase the diode turn off noise.
Then the bypass capacitors are ganged in parallel too close together in values, not to mention the series use of the 4700/63 main filters. As these are typically 20% parts during discharge the voltage across the capacitors almost certainly will not be equal. This will result in different discharge currents and additional non-linear reactance of the power supply rails.
Then there is the electronic regulator which at least uses resistors to try and stabilize the capacitance balance issues, but then also does this to the bypass (some of which are excess) caps and that only serves to raise their inductance and reduce their effectiveness.
The zener diode is directly in parallel with the series regulator filter by using an isolation resistor or inductor you get more effective filtering.
Now the other real issue as I read it is that SW1 changes the overall gain of the amplifier. That path goes through connectors and who knows what in actual implementation. The switch and all the connectors add a small bit of contact distortion.
Later we can talk about the rest of the inadequacies.
But the power supply issues can be huge because you bench test with a clean line and the user rarely has one. So you can measure well and deliver ....
Last edited:
John,
The design needs minor tweaks (and maybe a major one ) and correct PCB layout, plus much less crippled power supplies.
But let's not kid ourselves, what really limited the product was handing off an advanced concept to a bunch of engineers who did not remotely understand the design for practical implementation and not retaining a "hard" veto on the parts substitutions and other "minor" design choices.
In such a case results are left to chance...
Ciao T
John Curl said:
The design needs minor tweaks (and maybe a major one ) and correct PCB layout, plus much less crippled power supplies.
But let's not kid ourselves, what really limited the product was handing off an advanced concept to a bunch of engineers who did not remotely understand the design for practical implementation and not retaining a "hard" veto on the parts substitutions and other "minor" design choices.
In such a case results are left to chance...
Ciao T
Not much really, and some of Harry's strongest fans hate the mention. More like Colin Nancarrow or even Antille.
Cage was more an intellectual insider mixing the newly popular eastern/Zen ideas and music, and then there is 4'33".
I appriciate Cage but Harry was more "outsider" art.
Last edited:
Right you are T. Everyone else, yes to some degree. Apparently only T has the DIRECT experience that it takes to understand the problem of working with other engineers.
To start:
Too many caps in the power supply line.
Slow diodes.
Wrong size and type of feedback resistor.
Leakage from the transformers, and their 'phasing'
Wiring.
Layout, but this was not as bad as we first thought.
Input connectors and output connectors (just use a magnet to note the difference).
Zener diode buffering to lower noise.
And much more that is not shown here from the schematic, but we have found, important.
To start:
Too many caps in the power supply line.
Slow diodes.
Wrong size and type of feedback resistor.
Leakage from the transformers, and their 'phasing'
Wiring.
Layout, but this was not as bad as we first thought.
Input connectors and output connectors (just use a magnet to note the difference).
Zener diode buffering to lower noise.
And much more that is not shown here from the schematic, but we have found, important.
Now, what about the 'imaginary' part? Well the HCA 3500 was advertised as a 'John Curl' design like the HCA 2200 Mk2 that preceded it. The HCA 3500, like the HCA 2200 was a very 'cost effective' good bang for the buck, component.
The biggest engineering difference between the HCA 2200 and the HCA 3500 was power output, with some minor changes in the schematic and parts.
Yet, from the SAME magazine, the HCA 2200 got a B rating (exactly what I hoped for) and the HCA 3500 got NO rating, although both were reviewed extensively.
WHY? Well, many engineers here would say that the 'reviewers' are fools, and they should have accepted both as OK. Others would say that we did not advertise enough.
Yet, this attitude is what makes or breaks quality audio designs, and why most electronic engineers are not the best at designing audio products.
The biggest engineering difference between the HCA 2200 and the HCA 3500 was power output, with some minor changes in the schematic and parts.
Yet, from the SAME magazine, the HCA 2200 got a B rating (exactly what I hoped for) and the HCA 3500 got NO rating, although both were reviewed extensively.
WHY? Well, many engineers here would say that the 'reviewers' are fools, and they should have accepted both as OK. Others would say that we did not advertise enough.
Yet, this attitude is what makes or breaks quality audio designs, and why most electronic engineers are not the best at designing audio products.
Last edited:
Now, HOW DID I KNOW, that the design had problems, and even in its price range, was not necessarily good enough to compete with similar products?
Well, first, I had two trusted colleagues LISTEN to the HCA3500 in their OWN listening system, after I personally tested the same unit, myself, with my own test equipment, and found that it worked just fine.
One colleague refused to take it to CES, even as a back-up amp, after listening.
The other colleague listened to the same unit, and gave me the news. THEN we made a serious and successful effort to 'improve' it. Ultimately, we made units for ourselves and others, that pushed the competition aside and we could happily use it along with the CTC Blowtorch at CES. From the outside or even through normal measurements, very little would be noted as a difference between a stock and a modified unit.
(Now I am going on like KBK) '-)
Well, first, I had two trusted colleagues LISTEN to the HCA3500 in their OWN listening system, after I personally tested the same unit, myself, with my own test equipment, and found that it worked just fine.
One colleague refused to take it to CES, even as a back-up amp, after listening.
The other colleague listened to the same unit, and gave me the news. THEN we made a serious and successful effort to 'improve' it. Ultimately, we made units for ourselves and others, that pushed the competition aside and we could happily use it along with the CTC Blowtorch at CES. From the outside or even through normal measurements, very little would be noted as a difference between a stock and a modified unit.
(Now I am going on like KBK) '-)
- Status
- Not open for further replies.