Pass F5 DIY - help please :-)

[Sub Sonic]

Hi all,

I have been finally nearing completion on an F5 build, and have run across a few issues that now have me stumped. I am fine with speakers but amps are normally out of my area, so I’d appreciate any suggestions from the more knowledgeable/experienced members.

The amp is a First Watt F5 clone, with boards and on board components supplied by Jims Audio (ebay). The boards look very nice and I believe the name brand parts are genuine, despite some doubts earlier on. All transistors came as matched pairs. The amp design has no caps, the caps visible on the boards are simply across the incoming supply. The build is dual mono, with a 300VA toroidal T/F supplying each channel, giving a rail voltage of +-24VDC.

The power supply was tested separately before powering up the modules, it was powered up slowly on a Variac, and no fuses were blown. Rails are balanced within around 0.1V.

Everything went fine until it came to setting the bias. Despite allowing the unit to warm up for 1-2 hours and playing with the bias, I cannot get the bias reasonably balanced with minimal DC offset. The spec is for 0.59V across the 0.47 ohm R11/R12, but the DC offset is around 110mV for the left channel and around 600mV (!) for the right. It is not possible to tweak the bias enough on the +/- side to get this to an acceptable, reasonably balanced level.

I have measured all resistors in circuit, comparing the +- sides and also between channels and all appears pretty much identical.

Can anyone suggest what to try next?

 

Cheers,

 

SS

[almikel]

Hi SS

First suggestion is to re-post on the DIY forums that Papa frequents, and others that have built F5s (you've probably already done this?).

 

It certainly appears that something is "wrong" for them to be that far out.

 

I'm not used to seeing single sided boards with any solder on the component side - but it's also inconsistent.

Why is there any solder on the component side of the board?

Look at the leads on the ZTX450.

 

There may be a dry joint somewhere?

 

Good luck - I'm sure you'll be very happy once sorted.

Hopefully others much cleverer than I will chime in.

 

cheers

Mike

 

[almikel]

very nice build by the way - it will sound fabulous very soon I'm sure 

 

mike

[Sub Sonic]

Thanks Mike.

I have looked at the DIY Forums, there is a huge amount of info to wade through there :-)

The boards are through plated, hence the solder on both sides in some places. It does actually make it easier to install cables etc after the boards are mounted.

I'm pretty certain there are no dry joints, but anything is worth a check. I'll have a look at the transistors, it did occur to me that a transistor could be at fault, but I have no way of checking them.

Thanks for the suggestions,

Matt.

[Sub Sonic]

BTW I can't claim credit for the whole build - the amp chassis and soft start/bridges were existing, but I was more than happy to use them :-)

[Full Range]

On my F5 I purchased recently from another member It looks like the the authorised DIY audio boards were used

They look a little different to your PCBs but that could be the different parts used ( see photo )

@Green_Wagon did my bios settings and found it difficult so I hope he comes in to advise

However I have in the past month I have upgraded the F5 to the F6 and the bios setup is way easier to the F5

I also recommend what @@almikel has suggested

Edited December 27, 2015 by Full Range

[Sub Sonic]

Thanks @@Full Range. I'll have a look at the ZTX450/550. They are cheap enough to just replace anyway, so I might do that.

 

Out of curiosity, how would you compare the sound of the F5 to the F6?

 

Regards,

 

Matt.

[Full Range]

Thanks @@Full Range. I'll have a look at the ZTX450/550. They are cheap enough to just replace anyway, so I might do that. Out of curiosity, how would you compare the sound of the F5 to the F6? Regards, Matt.

They are different

The F5 is more forceful but in a nice way

The F6 is more exact but more analogue in nature, and better 3D presentation

Bass - is extended slightly more than the F5

The F6 however has a better tighter multi layer bass presentation with lots of inaudible bass vibration felt even at low volume

[mloutfie]

Ah the joy of setting up bias on Nelson Pass's design. Took me a while to get mine right. 600mV is really close to 0.59V so I wouldn't worry too much on that side if You are measure the voltage drop on R11/R12. DC offset is when you measure between the output + and - not voltage drop on a resistor. from my experience when a transistor is blown the voltage drop would be much higher than spec so don't think you have a blown transistor. If I were you I will recheck whether all the resistor used is the correct value and then do a continuity test to check whether any parts is not soldered properly

 

Note this is not a professional opinion just sharing my experience

[Green Wagon]

mmmmmm

 

yes, pass labs forum at diyaudio.

the threads can be, large. but contain decades of experience and fault fixes.

sometimes you just need to set a week or three aside and read.

I suggest skipping every other page as any fix etc will span several pages.

 

There are multiple versions of the f5 board. i dont believe it matters which you use.

 

yes biasing the f5 was a bundle of fun.

several days of fun.

And i still wasnt 100% happy with it.

 

main things to remember;

 

the offset is a result of the two bias pots.

it is slow to respond to adjustment.

tiny adjustments at a time. TINY.

three multi meters make life much easier but two are almost as good (I only had two. one bias point, one for offset)

'lid' on or off, it does make a change. Papa says lid on. a sheet of cardboard is good enough.

 

if you cant get the bias right with a low offset, try 'reversing' what your doing.

that is, if the bias ok but the offset is high and you cant get enough range from the needed pot, then back the bias off a little, and try again.

it can take half hour or more for it to thermally stabilise and have the voltage settle down.

 

I got pauls f5 to a stage where even with 20 turn pots i was still struggling to get it close with turns of the pots that were barely visible. as in barely turning the pot at all. 'Bumping' the set screw is a better description.

 

get the bias to roughly 0.5v and offset under a volt.

let it settle for an hour (if its already got heat in the heatsink) then do any adjustments in TINY steps.

but do keep an eye on it as it can start running away. (set for 0.5 bias, half hour later its sitting at 0.7 and running really hot)

as a final check, measure the heatsink temps. aim for around 20 degrees above ambient.

 

Pauls f6 is biased lower than recomended as his heatsinks cant handle much higher.

but f6 biasing isnt as 'sound altering' as the f5.

 

EDIT, oh, Pauls f6, it has the unofficial mods to zener and feed resistors as calculated by Zenmod.

Made the bias and dc setting a lot more stable and accurate.

Edited December 27, 2015 by Green Wagon

[Sub Sonic]

Ah the joy of setting up bias on Nelson Pass's design. Took me a while to get mine right. 600mV is really close to 0.59V so I wouldn't worry too much on that side if You are measure the voltage drop on R11/R12. DC offset is when you measure between the output + and - not voltage drop on a resistor. from my experience when a transistor is blown the voltage drop would be much higher than spec so don't think you have a blown transistor. If I were you I will recheck whether all the resistor used is the correct value and then do a continuity test to check whether any parts is not soldered properly

 

Note this is not a professional opinion just sharing my experience

Thanks for that. I should have been clearer, the DC offset was measured at the speaker output. Bias was measured across the 0.47 ohm resistors. I just can't seem to tweak the bias pots enough to get the bias balance and the DC offset happy at the same time.

Cheers,

Matt

[Sub Sonic]

mmmmmm

 

yes, pass labs forum at diyaudio.

the threads can be, large. but contain decades of experience and fault fixes.

sometimes you just need to set a week or three aside and read.

I suggest skipping every other page as any fix etc will span several pages.

 

There are multiple versions of the f5 board. i dont believe it matters which you use.

 

yes biasing the f5 was a bundle of fun.

several days of fun.

And i still wasnt 100% happy with it.

 

main things to remember;

 

the offset is a result of the two bias pots.

it is slow to respond to adjustment.

tiny adjustments at a time. TINY.

three multi meters make life much easier but two are almost as good (I only had two. one bias point, one for offset)

'lid' on or off, it does make a change. Papa says lid on. a sheet of cardboard is good enough.

 

if you cant get the bias right with a low offset, try 'reversing' what your doing.

that is, if the bias ok but the offset is high and you cant get enough range from the needed pot, then back the bias off a little, and try again.

it can take half hour or more for it to thermally stabilise and have the voltage settle down.

 

I got pauls f5 to a stage where even with 20 turn pots i was still struggling to get it close with turns of the pots that were barely visible. as in barely turning the pot at all. 'Bumping' the set screw is a better description.

 

get the bias to roughly 0.5v and offset under a volt.

let it settle for an hour (if its already got heat in the heatsink) then do any adjustments in TINY steps.

but do keep an eye on it as it can start running away. (set for 0.5 bias, half hour later its sitting at 0.7 and running really hot)

as a final check, measure the heatsink temps. aim for around 20 degrees above ambient.

 

Pauls f6 is biased lower than recomended as his heatsinks cant handle much higher.

but f6 biasing isnt as 'sound altering' as the f5.

 

EDIT, oh, Pauls f6, it has the unofficial mods to zener and feed resistors as calculated by Zenmod.

Made the bias and dc setting a lot more stable and accurate.

Thanks Nigel, that's useful information indeed.

I can definitely see the benefit in using 3 meters, it did occur to me. I think I might actually do it now :-) I'm going to bring the amp home from work (currently on holidays) so I can let it stabilise for longer rather than just having a go for a couple of hours at a time. I might even grab the IR thermometer to keep an eye on the output device temps too. Luckily the heatsinks seem more than adequate, at around the recommended bias level the sinks are fairly warm but not too hot to hold continuously.

Thanks again!

Matt

[Decky]

You have to be patient with bias adjustment as posted earlier. All Class A amps require very long worm up before you even start touching those pots. There was a difference with and without lid on when I was doing mine. DIY Audio is a stressful hobby - go easy on yourself ...

[Green Wagon]

The room in which I do repair work, theres one window and one door.

For biasing the window stays shut, the fan off, and the lid drops on.

 

Yeah, Class A, Extended Idle, No air flow, It gets warm in that little room.......

 

I always thought those that recommended having the lid on were nuts.

But, it can make a huge difference.

 

I started Pauls as soon as i got up (7ish), did tweeks each hour, and did final checks before bed (11ish).

For several days.

 

Others seem to adjust their f5 with ease, for me it was a pig.

The f6 (with mod) was so much quicker and more friendly to adjust.

 

No matter what your adjusting, unless it's way above the set point, then small steps it is.

[Decky]

Change the output devices.

[Sub Sonic]

Change the output devices.

Why's that? Can they change characteristics, or would they normally completely fail?

Cheers,

SS

[Full Range]

I believe that @@Decky built the F5 I purchased

Now that I have upgraded to the F6 I will Probably sell the populated F5 PCBs soon

[Sub Sonic]

PM with a price please? :-) I'll have a crack at this one first though.

[Decky]

Why's that? Can they change characteristics, or would they normally completely fail?

Cheers,

SS

 

There are so many fake transistors around. They work but no guaranteed properties of might be completely different device with a fake label that sells for more money. 

[Sub Sonic]

Cheers Decky. Do you know if matching is required for the output transistors? If so, I might have to buy a few and ask a mate nicely...

 

Regards,

 

SS

[Green Wagon]

I believe that @@Decky built the F5 I purchased

Now that I have upgraded to the F6 I will Probably sell the populated F5 PCBs soon

 

 

I believe your right in that Decky built those f5's.

From what I know he uses pretty good parts then tweeks further..

Could be a good way for someone to get into an f5 without the time and effort of sourcing the harder to find parts.

Plus there basically ready to run.

Bias will be a tad lower than 'recommended'.

[Sub Sonic]

Ok, things are improving. The 3 multimeter method is definitely the way to go. The right channel is looking good, and holding good bias and offset reliably. The left channel still has a few issues but some progress has been made. I have found that moving thermistor TH1 away from the main transistor Q4 has allowed adjustment, but the offset is at the top end of the expected range, around 50mv. Could it be a faulty or out of range thermistor?

Regards,

SS

[Green Wagon]

It helps when you can see what the other hand is doing

 

Dont normally see dud thermistors.

The thermistor is part of the biasing system, it's across the bias pot so should only be giving small corrections with thermal tracking of that output.

I think they claim that the best spot for it is on pin 2 of the fet (center leg) but thats easier said than done.

Also a little risky i think..........

next best, it must touch the fet body for reliable tracking.

 

If the output fet is not coupled to the heatsink to well, it will be getting hotter than it should, that will upset the thermistor and the biasing.

 

Looking at your pic, I note it looks like youve used silpads and thermal grease (fair wack of grease too).

If thats the case, that could be part of your problems.

 

on the various 'help me' threads on diy, those that have biasing issues sometimes have heatsinking issues.

These amps are fussy about a really good thermal interface with the heatsink.

 

People are using silpads as micas are getting harder to find.

Lots of posts about people getting large device micas, cutting them down, and even splitting them thinner.

 

i would suggest if you are using silpad and grease, strip that module out, clean of all grease, and reseat the outputs with a new silpad (no grease - thats what the silpad is spossed to do)

careful with tightening torque too. to much and it wont sit flat. thats why they use the large washer. copying Papa

 

These outputs get pushed pretty hard, that heatsink interface can make or break it.

 

Hey, you said you had a laser probe ?

See what the fets are running at ?

wouldnt be suprised if the temps are different on the naughty channel.......

 

edit.

the offset is a function of the difference of the two bias pots.

thats why it takes forever to set..

change one pot, it changes the bias of that fet, so it needs to thermally settle down at its new temperature.

 

50mv isnt bad. way better than most amps i see......

thats 0.05vdc 

yes id try to get it lower just because thats what im like.

chances are that might be the best it can achieve.

Edited December 28, 2015 by Green Wagon

[Equaliser]

Bias them once the amp is warmed up (30 minutes or so) then power it down and wait 5 minutes, power it on again and measure the bias and readjust if needed. If it's drifting completely out of spec then something else is at play. There's a huge thread/guide on diyaudio by poster 6L6 that can help you iron out some problems you may have. The F5 I built half a moon ago had heat problems in the small enclosure so certain resistors was going out of its nominal specification zone. Did it sound good but ran hot.

[Sub Sonic]

Thanks GW and DQ.

The thermal washers I used are these:

http://au.rs-online.com/web/p/thermal-gap-pads/7073364/

They are silicone coated Kapton, definitely thinner than the usual silpads. I did wonder about transfer paste but they felt so thin I added a thin smear. When I picked the amp up from work today I forgot the IR thermometer but I think I have one somewhere here from radio controlled car days, will have a look...

SS