F5 Turbo Builders Thread

Well, I found the short. I must have tightened the transformer bolt too tight, as there was a tear in the wrapper that exposed the copper to the chassis. Once I insulated the tear, the amp passed the lightbulb test and I was able to set the bias. The 2200 ohm r5/r6 resistor change appears to have been spot on. I was able to set initial bias 220 ohms with plenty of turns left on the pots, and it's cooking now. After two hours I'll up it to 320.
 
Hum Breaking Resistor (HBR) Location

Alright, I've successfully set the bias to 320mV on the amplifier and it's working...for the most part!

With inputs disconnected from a source, the amp does appear to have a very slight 60 hz hum. It's only noticeable if I put my ear right up next to the drivers on my speakers, and even then, I need to listen for it. Frankly, it's something I can live with and it's not something I'm all that worried about.

When I connect a source to the inputs, however, I'm getting a buzz that's more audible. I don't believe it's an AC ground loop issue because the buzz occurs even when when I connect the amp directly to a battery-powered source (i.e. phone, laptop). The buzz is also consistent and doesn't rise or decrease in level when I change the volume on the source, so I don't believe it's a signal interference issue.

I did some reading in the F5 thread and came across this post regarding cross-channel ground loops and how to minimize them with a hum breaking resistor (HBR) on the F5 boards:

http://www.diyaudio.com/forums/pass-labs/121228-f5-power-amplifier-1534.html#post4938077

I think the cross-channel ground loop may be my issue and I would like to try adding the HBR to my F5T 3.0 boards. I believe the correct location for the resistor would be between the input ground and the link pad shown in this photo from the build guide:

IMG_2280.jpg


Is this correct?
 
I went ahead and added 2 ohms of resistance (3W rated) at the location shown above, with the signal ground connected to the Link pad. It completely eliminated all hum and buzz! The amp now sounds phenomenal. The jfets from punkydawgs are the real deal, although they did require the R5/R6 resistor change to set the bias correctly.

I'm going to go ahead and call this project complete.
 
ground loop noise and class a power.

Thanks Kyle, I've been chasing that noise problem for some time, did the same, removed the link and put a 2.2 ohms, 2 Watts resistors. Now it's dead quiet even using non shielded interconnects.
I have a question for the guys who know.. While outside the system, on the bench I connected I function generator to the amp through a preamp. It clipped at 27 volts end into a 6.5 ohms resistor, that is 112 Watts. This is a V2 running at +/- 43 volts at the PS output with load, devices are biased at 0.350 you volts=0.7 amps so total bias is 2.8 amps. From these numbers, how do I get to the power number where the amp leaves class A to become class AB? In other words what is the class A output of the amplifier.
Thanks for your kind answers.
 
Leaving class A

Nelson says in that article: "Push-pull amplifiers generally operate in Class A mode up to a point where the output current is twice the value of the bias current. In the Class A region, both halves of the circuit share the signal simultaneously. Beyond that the signal is handled solely by the push (+) half of the amplifier or the pull (-) half."

Am i correct thinking that if the output transistors have 0.7 A rms of bias, then they will stay in class up to twice that number? I=1.4 A rms
Somewhere in the forum I read that there is usually a loss of up to 5 volts in the amplification stage. The rail voltages being +/- 43, minus the voltage drop: 38 Volts x 1.4 amps = 53.2 watts of class A.
Is this correct?
 
Thanks Kyle, I've been chasing that noise problem for some time, did the same, removed the link and put a 2.2 ohms, 2 Watts resistors. Now it's dead quiet even using non shielded interconnects.
I have a question for the guys who know.. While outside the system, on the bench I connected I function generator to the amp through a preamp. It clipped at 27 volts end into a 6.5 ohms resistor, that is 112 Watts. This is a V2 running at +/- 43 volts at the PS output with load, devices are biased at 0.350 you volts=0.7 amps so total bias is 2.8 amps. From these numbers, how do I get to the power number where the amp leaves class A to become class AB? In other words what is the class A output of the amplifier.
Thanks for your kind answers.

You say this is a v2. so 2 output pairs? in that case you have 1.4A bias. around 30W class A @8ohm. and 15W class A @4ohm.
 
I noticed how after the Bridge rectifier the voltage would go up x 1.44

I have a chance to get a 1000VA transformer but after rectifier it looks like it might be around 38V, way too high for standard F5 which id planned on building but perhaps ok for Turbo V2
Is there anything that can naturally lower the voltage if say my transformers voltage was higher than I wish ?
how about a regulator ? or what about an inductor ? I noticed the voltage will also go down when PSU under load.
 
Last edited:
I noticed how after the Bridge rectifier the voltage would go up x 1.44

I have a chance to get a 1000VA transformer but after rectifier it looks like it might be around 38V, way too high for standard F5 which id planned on building but perhaps ok for Turbo V2
Is there anything that can naturally lower the voltage if say my transformers voltage was higher than I wish ?
how about a regulator ? or what about an inductor ? I noticed the voltage will also go down when PSU under load.

I'd say cascode the input jfets and build a V2. If you have the diyayduo store boards it's easy to do. May sure your PS has 50 V capacitors.
 
I noticed how after the Bridge rectifier the voltage would go up x 1.44

I have a chance to get a 1000VA transformer but after rectifier it looks like it might be around 38V, way too high for standard F5 which id planned on building but perhaps ok for Turbo V2
Is there anything that can naturally lower the voltage if say my transformers voltage was higher than I wish ?
how about a regulator ? or what about an inductor ? I noticed the voltage will also go down when PSU under load.

A C multiplier "regulator" will drop around 4 volts and typically makes a very good PSU... There's one here Juma's Easy-Peasy Capacitance Multiplier
 
Pop pop pop pop

Had one of my v3 monos on the bench for some power supply upgrades. Well, I blew it up doing some testing under load. I changed the frequency but forgot to turn down the generator's output first. The amp outputs popped like a zipper and let out the magic smoke. :censored: Now to match up another set of 8 outputs and hope I got lucky all 4 jfets survived!

After the upgrades I was attempting to investigate an issue I've had from when the mono blocks were one stereo v2. If I turn it on and start listening cold before they fully heat up it sounds clean and smooth, really really good. There's absolutely no hum and can barely hear hiss with my ear against the tweeter. Super quiet! The issue starts after they heat up (max 51c depending on ambient). It develops a distortion. It goes from a clean & smooth sound to a kind of harshness in the upper mids & highs. It's not a full on ear assault, it's subtle, but enough to want to shorten the listening session. I wasn't able to see any oscillations on a 5mhz scope.

The important stuff:
I'm using the diyAudio store boards for everything. P3 is set in the center (measured). The cascode divider resistors are 12k (R25 & 26) & 4.75k (R27 & 28) and measure ~12.2v at emitter. Jfet current is 12.2ma||pair **there's two LSK170-BL and two LSJ74-BL per channel so 6.1ma per jfet. Rails are at +/- 46.5v (fully biased to 700ma/pair), CRCRC (30k+30k+60k uf per rail, 240k uf per mono). The bias resistors (R5 & 6) had to be changed from 1k to 2.2k to get enough range on the bias pots. The mosfet gate resistors were increased to 470R. I have 1nf caps in C3 & 4. All other values are same as per Nelson's schematic below.

Have any other F5T builders encountered anything like this? Any ideas? The phenomena happened with 2 different preamps, BA-3 and Salas DCG-3.

I was also was wondering if when using the doubled up jfets should each one get it's own source resistor, like pearl2 & others, or is it ok to just use the one for each paralleled pair? Should I adjust the jfet source resistors to bring the current back up to ~8ma each?

Sorry for the long post. I wanted to get all the details in.


F5Tv3_schematic.jpg
 
I have a F5tv2 that also seems to sound clean and smooth when cold. Listening to the amp at full operating temperature fatigues me a bit. I know it goes against what were told to believe about warming up equipment before any critical listening. I wish I had equipment to measure what my ears hear but don't. The problem may be Pshycho Acoustical Brain Fatigue Disorder or maybe overheating input JFETs. I have recently clamped some largish copper heat sinks on the FETs and think it may sound better. Does anyone know if JFETs distort with rising temperature?
 
I read that P3 can change the sound from most 2nd harmonic distortion to more 3rd harmonic distortion. Could it be that there is more 2nd harmonic when cold and more 3rd when warm?
When you say P3 is set to center and measured is that for P3 alone or for the whole circuit it should "balance"? .....if P3 alone is set to the center it may not be the "center" for the whole circuit?


How does the sound change when P3 is adjusted when the amp is warm?
 
I doubt it's an overheating issue but it could quite possibly be a distortion profile issue. I haven't messed with P3 yet and my test amp is under repair atm. They were set to equal resistances without power as instructed in the manual & build threads. Something to look at when I get it back together for sure.


That still leaves the question on the jfet source resistors and their bias when using the dual option. Should each jfet get it's own source resistor? If not then should the source resistor be lowered to bring the shared current back up?


Another thing I noticed comparing the F5 & F5T is the thermistor arrangement. I don't know if it's been asked before but in the F5 it's strung directly between the output gates and their respective rails. The F5T has them between the gates and source resistors. I wonder if it was a drawing mistake or was it changed to improve it? I don't think it would make a lot of difference but I'm curious.
 
Moderator
Joined 2002
Paid Member
If you are doubling Jfets with a single tail, the effective ratio of 1/gm and Rsource basically doubles (assuming properly matched JFETs). The effect of lowering this resistor (or using one for each JFET, which is exactly the same thing) and increasing the current is basically zero as the drain resistance for the same bias voltage to the outputs will also come down.

You may get a bit more current to drive a few more gates, though I've used V-grade Toshibas running ~9.5mA to drive 4 pairs without issues. When adding more pairs you only need to watch the input capacitance at both ends. Transconductance increases quite a bit after the first two output pairs, and the current required from the JFETs does not scale in a linear manner.

It is surprising that 12.2mA cannot generate enough bias voltage with the standard arrangement. With the thermistor at 500 ohms (assuming maximum temperature) and the bias pot fully open you should be able to generate over 7V of bias per rail. That sounds like a lot more range than needed, so maybe there are other issues with your build?

The thermistors were a worry for me as well, so I reverted to the standard arrangement used in the original, with values appropriate to multiple pairs. I've also used a 2-pair arrangement but given each device its own thermistor (thermally coupled to the drain of the device) with slightly better sonic performance, but it's a not a big difference. My issue was that the thermistor effectively controls only one of the output devices when connected this way though Nelson did explain that given the same temperature on the same heatsink, the devices equalise anyway.

The distortion you hear may be the sound changing after warm-up, the amp does open up quite bit after an hour or so. Unless you null the harmonic you're not going to know for sure though.