F5X -- the EUVL Approach
There are 100 ways to skin a cat. I made a promised last year to tell you how I would skin the F5 balanced circuit. In order to help me to focus on this task alone, I have opened a new thread to concentrate information relating to this approach only.
I shall try my best to answer all technical questions related to this approach (only) as much as I can, and as time allows. Please kindly bear with me if I choose to answer certain questions at a later date, e.g. after the prototype build. Please also kindly respect my wish to restrict posts to only those with technical content. If you wish to discuss other approaches, or to comment on personalities, or else, there are already quite a few other threads around for those purposes.
You can find further information about the construction here :
So where do we stand today and how we intend to proceed :
Thanks to the tremendous efforts of fitzfish and ashaw, we shall have the layout work for the following PCBs ready towards the end of February. The number in bracket represents the number of boards you would need to build one stereo amplifier (2 channels in one housing).
2 sets F5X amplifier board (main & daughter board)
8 pcs Capacitor Board *
2 pcs Regulator Board *
1 pcs Customised Control & Protection Board *
Option Stand-alone Control & Protection Board
Boards with a * are only required if you wish to follow the default solution to the last detail, including regulators, mechanics and case. If you wish to use your own solution for power supply, case, heat sink, etc., you would only need 2 sets of the amplifier boards.
To proceed further, the followings need to take place :
a. DIYA has to make a stand whether they wish to take over the organisation and the financing of the PCB production, including prototyping, once the intellectual work (PCB layout & associated documentation) is completed. I hope Variac or Jason would make a clear stand to this. This was at least the original idea.
If they no longer wish to be involved, then the team will decide whether and who would wish to take over that task. The two guys who did the work have naturally first claim to the rights after DIYA.
If no one in the team wish to take over, then I shall do this as a one-time GB, with the GB subscribers contributing to the extra costs of the PCB prototyping equally upfront. I emphasis that this will then be a one-time action only. I have no financial interest in this, and definitely not the time. So this is the least preferred scenario, probably for everyone but especially for me.
b. Once the prototyping PCBs are available, examples will be build, preferably by the two gentlemen who did the layout, to iron out any layout errors (the circuitries have already been proven by me in multiple builds). Once that is done, the PCBs are proven functional, and we shall publish all PCB related information, such as Bill of Material, component placement, overall dimensions, etc. If you wish to proceed to component sourcing at this stage, do bear in mind that the circuit is still not fully optimised, and that there might still be minor component changes as we fine tune the circuit further. This applies essentially only to the amplifier board.
It is very likely that we shall also be able to publish performance measurements (frequency response, square wave response, distortion spectrum, etc.) of the baseline circuit, as I built it a couple of years back.
c. As a further quest to make full use of the 4 quadrants of the F5X, test will be performed to deliberately bring the top and bottom halves of the circuit out of balance, with an aim to minimise 3rd and 5th harmonics. The correspondingly increased 2nd, 4th, & 6th harmonics can then be canceled between the left and the right halves. This may well be the most frustrating part of the test. Simulations has shown that if the devices are well matched in both Vgs and transconductance, the 3rd harmonics is already at a minimum with the standard circuit, and any deviation from this full symmetry not only increases second harmonics by orders of magnitude, but also increases (rather than decreases as we might hope) 3rd and 5th.
See also :
Distortion in JFET input stage circuits
d. The third stage of the development is to compare the cascoded version with the one without, to see how the distortion spectrum and the sonic impression compare. Simulations show little or no differences.
e. As a last variant, I wish to change the output devices to UHC-MOSFETs such as 2SK3497 / 2SJ618, and see whether they bring about any further improvements.
We shall publish results according to the 4 development stages as listed above. You can decide for yourself which is the best point to start building your own example. It is not unlikely that stage (b) is already the optimum with well matched devices. The aim is to at least finish step (d) by the end of June. But everyone in the team has a day-time job, so it is all on voluntary basis. Please try to show your understanding if we are behind schedule.
This concludes all the work that we shall perform, and the test team (AR2, fitzfish, ashaw) will write up, publish, and answer questions, as we go along, time allowing.
Dual-Mono Amplifier Case using a pair of Conrad MF35-151.5
As already mentioned earlier elsewhere, priority given to daytime business (quite rightly) has delayed the delivery of prototype parts for the case by some 2 months. We expect them to come towards end of February. Once they are checked for quality, we shall proceed to a mock-up assembly and then publish photos.
The same organisational issues of the PCB also apply here. The DIYA forum needs to make a stand how they want to proceed. The choice is perhaps again whether to order from their inline store, with them keeping stock, or (perhaps more likely) a GB, with an optional donation to the forum.
We already have rough price indications, but this of course depends on the batch size.
Some of us have already ordered prototype transformers from SumR.
And I have also published a transformer specification for alternative sources (e.g. in Europe) :
Fitzfish has kindly organised the purchase of the SumR Transformers. You may contact him directly for further details such as SumR model number, etc.
There are three known sources of Toshiba devices for the F5. At least two of them offer matched devices. So I shall leave you to organise those yourselves. We shall most probably publish later a simple method of matching transconductance (in additional to Vgs) for those who wish to achieve the ultimate performance.
I have also already published an article on matching the MOSFET source resistors. This applies equally to Fukushima MPC74’s as well as TO220 devices such as Caddock MP9xx series. Again there are 100 ways to skin a cat, so this is only one example method.
The reports will provide details of the requirements for the other important components, such as the feedback resistor network. But these are relatively straight forward and do not require extra attention at this early stage.
This concludes the first round of information I wish to release at this stage.
You are welcome.
As mentioned, we need the forum to make a few decisions before we can move forward.
Thanks for all the work to the whole team! Developing amps is a time consuming job, even more so when it's done properly ;)
A nice clean slate.
well i have to say although i thought it would be divisive, it does seem like the best idea and thankyou for the detailed writeup.
Patrick: i have a question on the specifics of the sumR transformer. i sent him the spec sheet you wrote up for him and he came back with a couple of questions that i hope you can answer, as i assume you already had the same questions from him for the build team.
b) (you have asked for the seondaries to be bifilar?). I would prefer to see some insulation between the secondary windings ( ie 500V hi pot )
c) Do you require Class B lead wires or will Class A lead wires be ok ? Or do you only require insulation to Class B? (our insulation materials presently meet Class B requirements).
i sent this to your email but it was with the payment info for the mosfets, so perhaps you missed it. with answers to these johnmclean and i can order our iron, it will take about 6 weeks for production and delivery over here, so will be nice to get it underway. i also have tx for my ajx and dcb1 with it.
|All times are GMT. The time now is 01:55 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright ©1999-2015 diyAudio