Y.A.F5Tv3.B.T. Yet another F5Tv3 Build Thread

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In the end... i was able to finally start some tests.
All the parts was mounted correctly, but i am struggling to get the bias up to desired point.
I end with the travel on one pot, and able to get "only" 60mv across the Rs (1//1 = 0.5ohm). I was able to zero the DC offset.

a quick recap of the setup: is a V3 with sk170+sj74 as front end, cascoded (2sc4793+2sa1837).
4x irfp240 + 4x irfp9240 as output devices. The diodes are MUR3020WTG from Onsemi. The thermistor was a 4k7 ntc.

the psu is providing +/- 55v.

At jfet i have a Vds of ~9.7v (i have used 10k+2k2 as cascode resistor values), and, as per calculations, is correct.
Across the 10 ohm resistor to ground of input jfefts i have ~45mV.

I have previously tested the fets for Idss and i was able to get ~8mA with 10V.

I am missing something ? why i can't go up with the bias ?
Also, i have tested with 1Khz@0.8Vrms and I was able to get just 11Vrms@8ohm at the output... i don't know the gain of the amp, but i think is a little bit low.

Maybe input fets are fakes.... What are your opinions ?
I have red around that maybe i need to change the value of R5/6 from 1k to something higher.
 
Last edited:
Show me a good picture of the front and the top of the JFET's, then I can tell you if they are fake or not.

Here you are... :D
 

Attachments

  • 20171126_151917.jpg
    20171126_151917.jpg
    206.2 KB · Views: 115
  • 20171126_151958.jpg
    20171126_151958.jpg
    390.4 KB · Views: 114
  • 20171126_152029.jpg
    20171126_152029.jpg
    313.4 KB · Views: 113
I have 9.7v of Vds on both jFets and 45mV over the their source resistors.
60mV max (can't beyond that) over output mosfets Rs.

If you point me exactly where you need voltage check i will later re-measure for you.
Is that 45mV across 10r (one source resistor), or across 20r (across the series pair).

The Drain load is that group of resistors.
You need to measure across the whole group, i.e. from supply rail to cascode collectors.
 
Last edited:
Is that 45mV across 10r (one source resistor), or across 20r (across the series pair).

The Drain load is that group of resistors.
You need to measure across the whole group, i.e. from supply rail to cascode.

Across only a 10ohm resistor, not the whole group.

So i need to measure from +Ve/ (-Ve) and the drain of Jfets ?
 

Attachments

  • v3_voltages.PNG
    v3_voltages.PNG
    111.1 KB · Views: 114
45mV / 10r gives the jFET source current > 4.5mA
There is a small current passing through the 100r (half the 200VR).
The total drain and source current through the jFET cannot be more than 1.1*4.5mA = ~ 4.95mA

The maximum resistor feeding the drain is 1k||5k = 833r (your VR may be slightly low in value, did you measure them before fitting?)
The maximum voltage drop cannot be more than 4.95*833r <4.125V
That 4.125V is applied across the combination of Vgs output mosFET plus Vrs across the 1r source resistor.

The question then comes down to :
What current will pass through the mosFET when the Vgs = 4.125-Vrs?

It seems the Vgs is too low for the mosFETs.
The standard solution is to increase the drain load on the jFETs.
Try changing the 1k (R5 & R6) to 1k3
Reset P1 & P2 VR to zero ohms. Try re-biasing and see whether you can set the output bias to the value you want.
Simples and described dozens of times in the various F5 Threads and offshoots.
And measure your set up voltages and write them on your sch for future reference. You can also add on the bias currents, again for future reference.
 
Last edited:
45mV / 10r gives the jFET source current > 4.5mA
There is a small current passing through the 100r (half the 200VR).
The total drain and source current through the jFET cannot be more than 1.1*4.5mA = ~ 4.95mA

The maximum resistor feeding the drain is 1k||5k = 833r (your VR may be slightly low in value, did you measure them before fitting?)
The maximum voltage drop cannot be more than 4.95*833r <4.125V
That 4.125V is applied across the combination of Vgs output mosFET plus Vrs across the 1r source resistor.

The question then comes down to :
What current will pass through the mosFET when the Vgs = 4.125-Vrs?

It seems the Vgs is too low for the mosFETs.
The standard solution is to increase the drain load on the jFETs.
Try changing the 1k (R5 & R6) to 1k3
Reset P1 & P2 VR to zero ohms. Try re-biasing and see whether you can set the output bias to the value you want.
Simples and described dozens of times in the various F5 Threads and offshoots.
And measure your set up voltages and write them on your sch for future reference. You can also add on the bias currents, again for future reference.

Thanks... the clarity of your explanation is very valuable !

I already knew that resistor could have been the culprint in this situation
by reading around similar threads... but i wanted to be sure that nothing else was misplaced.

A little by little I am getting about how this circuit works and i need to thanks all of you for your efforts in supporting my "endeavour".

At the moment I have swapped the 1k resistor for 3k3//3k3 (1k65).
now i am able to go high "as i want" with the biasing.
Right now is dialed to have 200mV across every Rs (1//1).
200mV over 0.5 = 400mA, with 55v = 22W of dissipation for each device...

400mA combined for 8 output devices give me 40W of classA power in 8 ohm.

The output offset is near 0.

I am testing with the same 1khz @ 0.8Vrms on the input and the output is still capped at 11Vrms over 8ohm.
I have seen that the original F5 has a gain of 15.3 dB... so is ~5x of voltage gain ratio. And if that is the case, i think the numbers, match (almost...).
Is that normal ?

And, another thing i am discovering, is that the bias is very "sensitive"... i can litterally blow over the fets and get the bias current to "jump" around a little bit...
 
Last edited:
check the value of the VR pots after you have set your bias and offset.
If the VR value is much lower than the maximum VR then that means it is passing more current.
You can correct for this by decreasing the parallel resistor (your two 3k3=1k65).
adding a 10k (=1k41) might be enough to get VR near to maximum, but you may have to go down 3k3 (=1k1)

For long term reliability you want the VR current to be low. Adding a fixed resistor in series with the VR allows the power in the VR to be reduced. Changing to a lower value VR+fixed resistor increases reliability even further.

Yes, these amplifiers are very sensitive to operating temperatures. That is why final setting up and checking, re-checking with the lid on is important.
 
Last edited:
Tonight i will check how the pots are set.
I will try to achieve the desired bias current by maximizing the VR resistance.

I have re-read (for the 100th time) the original f5turbo article and got that the gain is 22db, and that's gives me a x12.5 gain factor, and then the measured values matches.
(0.8Vrms in -> 11Vrms out).
In order to get a higher gain, i think the only way is to rise the feedback resistors (R7-10)... but i will investigate that more once i have the amp operating stable.
 
Last edited:
ok, with 1k65 (3k3//3k3) i had the pot @ 2k2.
now i have inserted a 1k1 (2k2//2k2) and now i can get ~150mV across Rs with the pot @ 5k.
Maybe i need to go to 1200 ohm to have a little bit more of headroom.
I am shooting for 3.5A of total bias, so i need to reach ~200mV across Rs.

As per preamp, what is a nice coupling with the F5.. i have lying around a BoZ from some years ago... do you think the combination is worth ?
 
ok, with 1k65 (3k3//3k3) i had the pot @ 2k2.
now i have inserted a 1k1 (2k2//2k2) and now i can get ~150mV across Rs with the pot @ 5k.
Maybe i need to go to 1200 ohm to have a little bit more of headroom.
I am shooting for 3.5A of total bias, so i need to reach ~200mV across Rs.

As per preamp, what is a nice coupling with the F5.. i have lying around a BoZ from some years ago... do you think the combination is worth ?
You went from 1k0 to 1k65 to 1k1
Use smaller steps instead of jumping like on a trampoline.
 
My sources are all a little bit lower, and i think to take advantage of the "new found muscle"-cit. i need a pre-stage with some gain.

My dac, for example, output 0.8Vrms.

As configured, the amp is biased with 3.5A total, the heatsink is ~45degC and the mosfet case, a little bit higher, around 50degc (case+ambient).

If i studied correctly your teachings, right now i have a 50ish Watt class-A amplifier, that can swing up to 45 volt (thus 140w AB).

If i have done my math correctly, i think with 22db of gain (as per papa article), is has a gain factor of 12. so with a 0.8Vrms sine input i will get ~ 10v of output, nothing more.
with 2Vac input, 24v output...


how many stupid things did I say?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.