Thanks ashaw, I'm interested if you find the bi-color LEDs.
"The difficult part for me for the protection board was crimping those 2mm connectors.
Had I known ashaw was using 2mm pitch, I would have asked for change back to 2.5mm.
Too late ...."
What would be suggested connectors to the PCB and cable assembly? The ones in the BOM could not be found anymore from Mouser.
"The difficult part for me for the protection board was crimping those 2mm connectors.
Had I known ashaw was using 2mm pitch, I would have asked for change back to 2.5mm.
Too late ...."
What would be suggested connectors to the PCB and cable assembly? The ones in the BOM could not be found anymore from Mouser.
Thanks for the tip. Here's a link for the Sherlock series in Digi-Key: Sherlock Connector System - Molex Connector | DigiKey
Digikey have the PCB headers https://www.digikey.hk/product-detail/en/jst-sales-america-inc/S4B-PH-K-S(LF)(SN)/455-1721-ND/926628
But again the mating cable parts seem to be only at Mouser 51065-0400 Molex | Mouser 香港
GB3 Updates
The F5X project has gone through 2 batches already with more than 50 examples in the field.
So GB3 is very matured and everything is basically the same as GB2.
There are a lot of information and discussions already in this build thread.
I have gone throught the thread again and picked some posts with particlular relevance.
But I still encourage you to go through the entire thread.
Post 661, 664, 738, 740, 823, 847, 933, 938, 948, 973, 977, 978, 983, 988, 1008, 1026, 1044, 1046, 1049, 1076, 1095, 1231
The GB3 subscribers will receive per Email an updated BoM in pdf format.
These are essentially the same as GB2, with the exception of some components which have become obsolete.
Patrick
The F5X project has gone through 2 batches already with more than 50 examples in the field.
So GB3 is very matured and everything is basically the same as GB2.
There are a lot of information and discussions already in this build thread.
I have gone throught the thread again and picked some posts with particlular relevance.
But I still encourage you to go through the entire thread.
Post 661, 664, 738, 740, 823, 847, 933, 938, 948, 973, 977, 978, 983, 988, 1008, 1026, 1044, 1046, 1049, 1076, 1095, 1231
The GB3 subscribers will receive per Email an updated BoM in pdf format.
These are essentially the same as GB2, with the exception of some components which have become obsolete.
Patrick
The most "difficult" of these are the power supply caps.
The original Panasonic 22000µF 35V is very compact at D30mm x H45mm.
It is also rated at 105°C, which is important in a warm housing.
Unfortunately it is no longer produced.
None of the readily available alternatives can fit 1:1.
So we have to make some choices.
Firstly why 35V ?
The cap has to survive the peak voltage after the rectifiers.
With 15Vrms secondaries and 15% no-load margin, this comes to ~23V
Too close for comfort to use 25V.
The closest I can find is Rubycon 35MXC22000MEFCSN30X50.
It is also 22000µF 35V, and slightly longer with 50mm.
But it is non-stock with a MoQ of 100 and delivery of 17 weeks.
MoQ is no issue with you group together to GB.
So it is an realistic option.
If you accept less capcitance (in theory not a big deal because of cap multipliers instead of CRC),
then Nichicon LGY1V153MELB45 (15000µF 35V) will fit, and is in stock.
You can of course ask the transformer supplier to limit no-load rectified voltage to 23V max.
Then you can use 25V caps with more choices.
Examples are EPCOS B41252A5229M000, or Nichicon LGY1E223MELB45.
If you are not worries about lifetime, but want to use Audio caps, then you can consider Nichicon LKG1V183MESBAK.
Non-stock, 200 MoQ, and 85°C.
I never like using caps to their full voltage with no margin.
So if I have to choose, I would stick to 35V, 105°C.
And then the choice would be between Rubycon MXC or MXG (non-stock), and Nichicon LGY.
Patrick
The original Panasonic 22000µF 35V is very compact at D30mm x H45mm.
It is also rated at 105°C, which is important in a warm housing.
Unfortunately it is no longer produced.
None of the readily available alternatives can fit 1:1.
So we have to make some choices.
Firstly why 35V ?
The cap has to survive the peak voltage after the rectifiers.
With 15Vrms secondaries and 15% no-load margin, this comes to ~23V
Too close for comfort to use 25V.
The closest I can find is Rubycon 35MXC22000MEFCSN30X50.
It is also 22000µF 35V, and slightly longer with 50mm.
But it is non-stock with a MoQ of 100 and delivery of 17 weeks.
MoQ is no issue with you group together to GB.
So it is an realistic option.
If you accept less capcitance (in theory not a big deal because of cap multipliers instead of CRC),
then Nichicon LGY1V153MELB45 (15000µF 35V) will fit, and is in stock.
You can of course ask the transformer supplier to limit no-load rectified voltage to 23V max.
Then you can use 25V caps with more choices.
Examples are EPCOS B41252A5229M000, or Nichicon LGY1E223MELB45.
If you are not worries about lifetime, but want to use Audio caps, then you can consider Nichicon LKG1V183MESBAK.
Non-stock, 200 MoQ, and 85°C.
I never like using caps to their full voltage with no margin.
So if I have to choose, I would stick to 35V, 105°C.
And then the choice would be between Rubycon MXC or MXG (non-stock), and Nichicon LGY.
Patrick
Another component that I have changed for my own monoblocks are the speaker relays.
Initially I have chosen high current automotive relays for lifetime reasons.
But these require a corresponding high minimum current for it to make electrical contact properly.
So in my J2 build, for example, I have tried out gold plated 8A relays :
Schrack RY212024 or RTE25024
The gold plating helps against oxidation and is worth having in any case.
FirstWatt J2
https://www.diyaudio.com/forums/att...477061-firstwatt-j2-191204-j2-k3497-proto-jpg
If you wish to try these, there is no change required in the protection circuit.
Just sibstitute. And you need a slightly different mechanical mounting, without saying.
Patrick
Initially I have chosen high current automotive relays for lifetime reasons.
But these require a corresponding high minimum current for it to make electrical contact properly.
So in my J2 build, for example, I have tried out gold plated 8A relays :
Schrack RY212024 or RTE25024
The gold plating helps against oxidation and is worth having in any case.
FirstWatt J2
https://www.diyaudio.com/forums/att...477061-firstwatt-j2-191204-j2-k3497-proto-jpg
If you wish to try these, there is no change required in the protection circuit.
Just sibstitute. And you need a slightly different mechanical mounting, without saying.
Patrick
F5X Monoblocks
I guess an update to the F5X project might provide some additional motivations to the GB3 subscribers.
The original prototype from 2012 is now with another XEN team member in Hong Kong.
So it is time that I rebuild a new pair, to go with my finished F5X Pre.
It is also known that I have a pair of Accuton CELL speakers with active crossovers.
Milestones Speaker project
The new F5X should therefore be adapted to drive the Accuton mid-range and tweeter.
Whereas the bass unit is to be driven by 3x LM4780 in parallel, configured as inverted balanced.
The difference in closed loop gain between the F5X and the LM4780 has to be taken care of in the active crossover.
This will be placed on the rear panel of the case.
So each case now accomodates the two amplifiers for one channel.
But it only has the dissipation of one F5X, plus a tiny bit extra from the chipamps.
Here is a picture of the finished amplifier modules.
More will come in the next weeks and months, as I finish the other modules.
Patrick
.
I guess an update to the F5X project might provide some additional motivations to the GB3 subscribers.
The original prototype from 2012 is now with another XEN team member in Hong Kong.
So it is time that I rebuild a new pair, to go with my finished F5X Pre.
It is also known that I have a pair of Accuton CELL speakers with active crossovers.
Milestones Speaker project
The new F5X should therefore be adapted to drive the Accuton mid-range and tweeter.
Whereas the bass unit is to be driven by 3x LM4780 in parallel, configured as inverted balanced.
The difference in closed loop gain between the F5X and the LM4780 has to be taken care of in the active crossover.
This will be placed on the rear panel of the case.
So each case now accomodates the two amplifiers for one channel.
But it only has the dissipation of one F5X, plus a tiny bit extra from the chipamps.
Here is a picture of the finished amplifier modules.
More will come in the next weeks and months, as I finish the other modules.
Patrick
.
Attachments
Nice build Patrick! Can't wait to see more.
I started to fill my Mouser shopping cart with the F5X power amp PCB parts in order to reach MOQ for free shipping and noticed that the RN55 33ohm and RN65/70 220ohm resistors are out of stock and available in June 2020. Is it ok to replace the 33ohm ones with Beyschlag MELF? There are pads for MELFs under the PCB.
How about Vishay CPF3220 for the 220ohm replacement?
Hificollective seems to have PRP 30 and 220ohm resistors in stock, but I guess they are out of the question. How bad were the reliability problems with these? Maybe they have improved.
I want the boards to look good as well, so I tend to use the same type throughout.
The pads at the back can be used, but they were primarily intended for trimming purposes.
(I might remove the trimmers later and replace with fixed Rs.)
And Mouser has plenty of 33.2R in stock.
Also CMF55 is essentially the same as RN55, so those can also be used.
In choosing resistors, I would be willing to spend a bit more and choose a lower tempco (e.g. 50ppm/K or lower).
It reduces distortion.
CPF3320 is not thin film.
CMF65 and CMF70 both have 250R, which you can parallel with a 1.8k (e.g. RN55) to get 220R.
I have not been using PRP lately. So I cannot comment.
But when we talked to them last, they had such a poor attitude to that defect, that we deciced not to deal with them anymore.
I rather trust Vishay for their quality control system, especially for MIL spec products.
Patrick
The pads at the back can be used, but they were primarily intended for trimming purposes.
(I might remove the trimmers later and replace with fixed Rs.)
And Mouser has plenty of 33.2R in stock.
Also CMF55 is essentially the same as RN55, so those can also be used.
In choosing resistors, I would be willing to spend a bit more and choose a lower tempco (e.g. 50ppm/K or lower).
It reduces distortion.
CPF3320 is not thin film.
CMF65 and CMF70 both have 250R, which you can parallel with a 1.8k (e.g. RN55) to get 220R.
I have not been using PRP lately. So I cannot comment.
But when we talked to them last, they had such a poor attitude to that defect, that we deciced not to deal with them anymore.
I rather trust Vishay for their quality control system, especially for MIL spec products.
Patrick
Some more progress in lock-down time.
As you can see, I do not follow the standard layout, which was decided by majority vote in GB1.
The aux. transformer supplies the active crossover and the protection board at the same time.
They are isolated from each other by separate 7812/7912 regulators.
The active crossover has additional local LT1761/1964 regulators.
Patrick
.
As you can see, I do not follow the standard layout, which was decided by majority vote in GB1.
The aux. transformer supplies the active crossover and the protection board at the same time.
They are isolated from each other by separate 7812/7912 regulators.
The active crossover has additional local LT1761/1964 regulators.
Patrick
.
Attachments
Super nice and tidy as allways Patrick.
I’ve made some progress too with the regular F5X. Waiting for brass standoffs to arrive before doing some testing with the power mosfets in place. Front end is tested now.
I’ve made some progress too with the regular F5X. Waiting for brass standoffs to arrive before doing some testing with the power mosfets in place. Front end is tested now.
Attachments
....Sorry... i am working on another version...should not have commented.
My version is prior to GB 3.....all is well with you.
I will follow GB3 better ;-)
Anyway input pair bias can be anything from 3 ma to 8ma....so plenty of latitude....
You can play later with bias of output pair...choices ..choices.
My version is prior to GB 3.....all is well with you.
I will follow GB3 better ;-)
Anyway input pair bias can be anything from 3 ma to 8ma....so plenty of latitude....
You can play later with bias of output pair...choices ..choices.
Last edited: