F5Turbo Illustrated Build Guide

6L6

Moderator
Paid Member
2010-10-22 6:43 pm
Denver, Colorado
F5Turbo v2 (with Cascode) Using the diyAudio F5T ver. 3 PCB

PCB available here - F-5T (6 PCBs, Makes 2 channels; Rev3.0)


Nelsons original article and schematics, please read this for the best information concerning this amp - http://www.firstwatt.com/pdf/art_f5_turbo.pdf

Schematic

Please Read — The PCB numbering has been rationalized to match the part numbers and placement on this schematic (The F5Tv3 from Nelson’s article)
You can build any version of the amp from this article, or have multiple outputs, use the Cascode or not, etc… But when referencing the silkscreen to the schematic, remember to use this one.
F5Tv3_schematic.jpg


Please have this printed out in front of you when you are stuffing the PCB.

There is one error in this schematic - R26 should be 10K

Also, in my opinion, C3 and C4, labeled here as ‘OPT’, are required for stability from destructive oscillation. This is a very high bandwidth design and those caps help protect it from ultrasonic HF oscillation. Use 1000pF. Also, if oscillation is a problem for your build, increase the value of the gate resistors (R13 -16 and similar) any value from 47.5R to 680R will work and sound fine, the higher the value the more damping occurs.

IMG_2274.jpg

A look at the new PCB.

The biggest changes are the increase in size of the PCB, in order to accommodate holes that will mount to the Universal Mounting Spec. Also there are multiple solder pads at each wire connection point, making it much easier to configure this to any of the F5T variants.

IMG_2280.jpg

If using single-ended (RCA) inputs remember to add this link at the labeled spot.

IMG_2275.jpg

Please remember that this build is showing the cascode circuit being used. It is not strictly necessary for 32V rails, but is required if you choose to use higher voltages. If building a v2, I would suggest not cascoding, simpler is better.

In this build the packages in R8 and R9 are not transistors but high wattage Caddock resistors, I used them as I had them on hand. Most builds will use 3W axial lead resistors in the R7-10 positions.
Also note that because the cascode circuit is used, there are capacitors in the C5, C6 positions. As well as all the associated resistors, and of course the transistors.

IMG_2278.jpg

This photo shows the Front-End board (hereafter referred to as “FE board” or “FE”) stuffed in the manner to build a v3 or v2 with cascode. The feedback resistors and cascode transistors have clip-on heatsinks, and the rest of the board is stuffed. The Q1.1 and Q2.1 position are left empty, they are there if you are going to make a huge F5T with more than 4 pair of output transistors, so you can parallel the Jfets and have more driver current.

Note regarding P3 - set to it’s center value before stuffing. Equal resistance from leg 1 to 2, and from leg 2 to 3.

IMG_1464.jpg



This is stuffed in the typical configuration for a F5Tv2, with the normal style resistors and all the cascode circuitry and transistors empty. (This photos shows the earlier version of the PCB.)

Regardless of the PCB version, if you don’t cascode you need to jumper collector to emitter pads on Q5 and Q6, shown below —

IMG_1466.jpg

Connect C to E for non-cascode. Watch the labeling, they are mirror image left to right.
(This photo shows the old version PCB, for this illustration there is no practical difference between the old and new.)

IMG_1456_zps2d4df6c5.jpg

(image shows the old version PCB, new version differs only in silkscreen labeling.)

The output boards are very straightforward, the thing to remember is to have the N channel on the right, which will make the hookups rational as well as placing the thermistor on the outside. Also take a few clipped-off resistor leads and make some terminals for the test points (TP1, 2) if you have a voltmeter with clippy probes.


IMG_1451_zps458bd3aa.jpg

Output boards mount as shown. Some people will suggest that the Mosfets should be lower than the diodes for most efficient heatsink usage. There may be a difference, I can’t tell much, I’ve tried it both ways. The photos with the v3 boards have the fets low.

IMG_2289.jpg

The output boards are shown here with slightly different resistors.Also note the standoffs in the center for mounting the front end PCB.

IMG_2290.jpg

The FE mounts in the center, and it’s easier if you attach some of the wires before installation.
In these photos the color coding is:
V+ = Red
V- = Black
P-Gate = Purple
Output = White
N-Gate - Blue
GND = Green (not shown in this photo)

The brighter green is the input signal, it’s coax and It will also show up in photos as bright red, depending on which channel I happened to photograph…

IMG_2340.jpg

Connections to the output boards.

IMG_2293.jpg


IMG_2294.jpg

The power from the PSU is connected to the FE board. It is routed as shown for the purpose of clarity and illustration. I strongly suggest you make these connections as short as comfortable.

IMG_2342.jpg

This shows the final connections, the speaker outputs, black and red down the center of the photo.


Power Supply

A stereo F-5T v2 requires a bipolar power supply of (+/- 32V). This will require an 24V+24V (or 48V Center Tapped) transformer from 600-800VA, and PSU capacitance of 80,000uF per rail or more. The F-5T v3 will require two of the above PSU. (One for each monoblock.) If you have the heatsink and proper voltage capacitors you may use even more voltage for more power, see the original article for more information.

Although not the exact PSU as shown in the Nelson Pass article, the diyAudio PSUv3 circuit board can be used to make a suitable power supply for the F5Turbo.

Using the PSUv3 board allows an easy addition of dual diode bridges, which will help keep the transformer mechanical noise to a minimum, and other benefits such as LEDs on each rail, room for (8) capacitors of up to 35mm diameter, the ability to use many different styles of connectors on the board itself, and not to mention the convenience of having everything on one PCB.

The diyAudio universal PSU v3 was made with the requirements of this very amp when it was designed.

An illustrated guide to building the diyAudio PSUv3 can be found here - diyAudio Power Supply Circuit Board v3 illustrated build guide

IMG_2282.jpg

This is an Antek 600VA 24V+24V transformer. AN-6224

IMG_2281.jpg

Showing the wiring of the AC to the transformer. This is for 120V

IMG_2283.jpg

With the PSU board stuffed and installed.

IMG_2284.jpg


Remember to test the PSU for proper voltages (which, with no load, will read a little high) before connecting any of the amplifier circuit. Meter set to DC volts, black probe on GND, with red probe read +32V on V+ and -32V on V- .


Chassis

This is a high-wattage Class-A amplifier. It has lots of excess heat that needs to be dealt with. The 5U Chassis is of suitable size and dissipation for a Stereo v2 or a monoblock v3.

The 5U and 4U Deluxe chassis from the DIY audio store are basically identical except for size and that the 5U heatsinks are 2 pieces per side.

An illustrated guide to building the Deluxe Chassis (4U) can be found here - Illustrated guide to the 4U deluxe Chassis

Testing and Powerup

Adjustments (Bias and Offset, set with P1 and P2)

This is easiest with three DMM.

Shorting the input jacks is helpful, although not strictly necessary.

Before power up, dial pots P1 and P2 to 0 ohms ( check with ohmmeter across TP1-2 and TP3-4 on FE board )

With voltmeter across PSU output ( best between + and - of PSU) observe max voltage of PSU - it never hurts to make sure it’s doing what you expect.

Place one voltmeter at output - to observe DC offset

Place voltmeters across TP of output board, one on N-channel, one on P channel.

For test - slowly dial up Variac ( presuming that you have one , as man with many skills) up to full mains voltage , observing rail voltage at PSU ....... thinking about max cap voltage ( 25V as in FW ? ) , because with 0 Iq PSU is unloaded and voltage is maxed (It’s useful to have another meter for this…) :)

If nothing is smells bad, and the magic smoke is still in the circuit - leave Variac at full mains ;

IF you don’t have a Variac, you must build a lightbulb mains lead.

DSCF0136.jpg


bulbtester.jpg


(see notes for mains lead below)

What's important - Iq (measured as the voltage across source resistors; the Mosfet bias) must be very low , offset is irrelevant in this moment .

Now turn one pot one turn ( assuming that you have multiturn pots) then turn other pot one turn. Continue, one turn at a time on each pot until something happens.

Observe voltage across resistors and output DC offset.

Proceed one then second pot , again just one turn

Observe Iq and offset

Again one turn + one turn

Now you are probably in range when you can see which pot is pulling offset in right direction - to 0 . It will feel like one of the pots is controlling the bias on both sides, and the other is controlling the DC offset.

It’s best to increase the bias a bit, and then zero the offset. As you zero the offset you will decrease some of the bias, so it will be two steps forward and one step back. That action is normal.

As you increase the bias and zero the offset, remember to always keep the offset near zero. If you run out of turn on the pots, determine your max bias, with zero offset. (It’s useful for troubleshooting)

Proceed iteratively with pots , while you set - say - 75% of desired bias, with zero offset.

Now - put lid on box and let it cook for a while - until you get thermal equilibrium on heatsinks

It's best to use wire/clips to leave those voltmeters in place ;

Open the lid , up bias to - say - 90% of desired one ,while maintaining offset

Put lid on , let it cook.

Check again.

If all is OK - move voltmeters for Bias and offset to other channel and repeat procedure.

Use it few days at 90% of desired bias , then check and set to 100%

Remember - temp. equilibrium with lid on is important

Setting P3

BEFORE installing and soldering P3 it’s best to adjust the pot so you have equal resistance from pin 1-2 and pin 2-3.

If you didn’t set it, determine how many turns the pot has. Run the pot all the way to one limit (they usually click) and then turn the adjustment the other way for 1/2 of it’s turns. (I.E.,if a 25-turn pot, adjust it 12.5 turns.)

Assuming well matched Jfets the neutral position is going to sound really nice, with 2nd harmonic dominant at the 1W level. IF you have access to a distortion analyzer, or a high-resolution FFT (or both…) give the amp a 1K test sine wave that outputs 1W into your load resistor. Then adjust P3 as necessary for the harmonics you prefer. If you adjust for minimum THD, you will likely have nulled out most of the 2nd harmonic and made it 3rd dominant, which in my opinion makes it very fast and clean, at the expense of soul. YMMV.



Lightbulb Mains lead notes -

If the bulb ever turn on and stays bright, you probably have a short.

Normal operation when turning on a cold amp will have the bulb glow bright for a second or two, then dim, perhaps off. (this is the capacitors charging, then full)

As you increase the bias of the amp the bulb will glow brighter, and this is linear with the bias amount. A fully biased amp can get the bulb to glow very bright.

You cannot set full bias with the bulb in place - as it increases the bulb will glow more, limiting the voltage to the amp and all the readings will be wrong compared to when the bulb lead is out.

You can, however, set the initial bias with the bulb in place. Start the procedure, turn the pots until something happens, and set, at maximum, 0.1V across the source resistors and zero offset. Getting the pots started this way is a good idea. Remember, this is with a reduced voltage, and the pots will make the circuit draw MUCH more bias when the normal mains lead in used. Expect to measure 0.2V or more with a normal cord. Continue biasing in small steps, always trying to keep the offset near zero.
 
Last edited:

6L6

Moderator
Paid Member
2010-10-22 6:43 pm
Denver, Colorado
Member chuckd wrote a wonderful bias procedure, here it is so it's easy to reference. :)

````````

Bias Procedure for F5 Turbo Version 2

After you have completed your build you need to bias the output (measured in mV) in order to generate enough power to activate the MOSFETS. You need to do this while keeping the DC Offset (the potential across the speaker terminals) at 0V. This is achieved by increasing gradually, in an alternating pattern, the resistance of P1 and P2. You are not directly measuring the resistance but the voltage across P1 and P2.

Things you’ll need:

3 x DMM
1x Variac
1 x completed F5 Turbo v2

Set P1 and P2 to 0Ω
Setting these to 0 prevents current flow through the FE stage while you power up the amplifier. To make sure you are starting at zero bias at the beginning.

- Attach leads to TP1 and TP2, and TP3 and TP4 on the front end board.
- Turn P1 and P2 whichever way is required to set the resistance across the test points to 0.
- Make an arrow on each pot to indicate what way to turn the pot to increase resistance. Clockwise may not be “up”.


Powerup the amp, check the PSU voltages.

- Make sure the Variac and the power entry module (PEM) (the Schurter switch) are correctly and properly fused.
- Connect the PEM to a Variac or light bulb tester with appropriately current rated power cord (AWG12 and above should suffice).
- Turn Variac to 0V.
- Switch Variac and amp on.
- Check voltage between positive and negative (neutral in the U.S.) with the DMM (set to check VAC) across the barrier block connection. Ensure it reads 0V.
- With the leads still connected to the terminals, slowly increase the voltage on the Variac. If smoke appears turn off the power and troubleshoot the problem. If the fuse blows, the there is a short in the system. Locate and troubleshoot the problem.
- If there is no smoke, continue slowly increasing until you reach mains voltage.
- Once at mains voltage, you can check the step down AV voltage across the secondaries to ensure it matches the value of the transformer.
- Switch DMM to measure DC volts.
- Check the voltage out of the rectifier and before the capacitor bank to ensure it is DC.
- Finally, check the DC voltage at the terminals of the PSU. This is the power that will go to the Gain Stage and Output boards.
- Connect the positive lead of the DMM to the V+ and the negative lead to GND. You should have a positive reading.
- Now connect the positive lead from the DMM to GND and the negative to V-. You should have the same potential but a negative signal.
- If you don’t you need to check your wiring.
- If you do then proceed to setting the bias.



Setting the bias

- Make sure the mains are at max voltage. Remove a light bulb tester if you are using one.
- Set your three DMMs to check DC voltage. If the DMM is manual, set to 2V or 20V.
- Using alligator clips, attach one DMM to TP2 and TP3 on the output boards: One on the P-Channel and one on the N-Channel. Ensure it reads 0V. These will measure the bias.
- Attach the third DMM across the speaker output. Ensure it reads 0V. This will measure the DC offset.
- Grab an insulated screw driver and turn P1 (or P2 – it doesn’t matter) 360o. There may be no change on the DMM. This is normal.
- Turn the other potentiometer 360o. Again, there may be no change in the voltage. This is normal.
- Continue alternating increasing P1 and P2 with 360o turns until you start to read a voltage across the potentiometers.
- While increasing the resistance, keep a watch on the DC offset. It will increase or decrease depending on which potentiometers you turn. It should remain around 0V.
- Continue alternating turning P1 and P2 and bring them up to 300 mV (0.3V). Ensure that the DC offset remains at 0V.
- As you get closer to 300mV, you may have to adjust the turns to a smaller amount since a full turn can have a dramatic change in voltage.
- The heat sinks should be warming up.
- Once you get the bias on each output board to around 300mV (there may be a slight difference in order to keep the DC offset at 0V. It is ok if the bias voltage difference is within 50mV, and the offset zeroed.
- Set the lid on and let the amp rest (or ‘cook’) for 20 minutes.
- Keep a watch on the DMMs. As the thermistors heat up they will adjust to keep the DC offset at 0V.
- The bias may change as the MOSFETs heat up. This is normal and should settle after about 20 minutes.
- After 20 minutes, and if the bias readings have settled, and there is no smoke, increase the bias to 350mV by alternating turning P1 and P2.
- You may not get exactly 350mV bias on both boards at the same time as keeping the DC offset at 0V. This is normal. Get the bias between P1 and P2 as close as you can to each other within the 350mV range all the while with DC offset at 0V.
- Let the amp cook for another 20 minutes.
- The bias should not change too much, if at all, since the heat sinks have already warmed up.
- If this is a stereo amp, repeat this process for the other channel.
- Once the other channel is done, recheck the bias and DC offset for the first channel, and make any adjustments to the bias to bring it up/down to around 350mV and 0V DC offset.
- That’s it! Congratulations, you have just biased your amp.
- Turn off the amp and let it cool a little.
- Attach your test speakers (if you have them) and turn on the power. If nothing happens then you should be able to integrate the amplifier to your system.
- If you don’t have test speakers, and you attach them to your system speakers, you run the risk of damaging them. So proceed down this path at your own risk.
 
Last edited:
Jim,

The resistors for the voltage dividers to the bases of the cascode transistors are in error on the schematic, as shown, in that they differ on each rail. I found all 4.7k or all 10k work fine with pre-load 32 volt rails (or close to 32V).

Steve
R25:R27 seem to be in the correct ratio and give ~9V Vds on the upper jFET.
R26:R28 do seem odd. They end up giving <3Vds on the lower jFET.

Equal values (1:1 ratio) will give >15Vds on the jFETs.
 
Last edited:
regarding R25-28. this is depending on rail voltage and desired J-fet voltage.
if they all are 4.7K or all are 10K. the result will be the same. about 1/2 of the rail voltage.
calculation will be:
rail voltage/(R26+R28)*R28=cascode voltage.

AudioSan,

What is the ideal or allowable range of voltages to the base of the cascoding transistors. That is, using a volt meter, what voltage would we like to see at the base of these devices after making in through the voltage divider and R29 or R30? I have read 9 volts, 15 volts, 17 volts. I'm sure there is a range, but I wasn't sure what is generally recommended. Thanks in advance.
 

Smittys

New member
2013-01-08 10:52 pm
6l6
Cool!! I'm in!!
So if I understand this correctly...for 100w class A I need 36 volts but I saw a antek trfo with 4x 35 volts is this enough? I will also cascade the jfets. Caps at 63v? I'm not sure about the bias resistors? Did I miss anything else?
Any suggestions would be appreciated!
Thanks
 
Class A is mainly about current capability but you also need enough voltage to push that current into the load. Follow AndrewT's numbers for current and work backwards to the voltage peaks that you'll need. Add a few volts for the amp's inability to swing rail to rail. With typical CRC power supplies we build you can expect rail voltage to be somewhere around 1.2x the transformer's AC output.

You also need to consider the transformer's power rating. For class A operation the VA rating should be at least 2x the amp's power draw (rail to rail voltage x bias current). Better yet is 3x or more to minimize heating and chance of mechanical hum.

Assuming you will end up with roughly 40v rails 63v caps are fine. The usual rule of thumb around here is add 20% to the rail voltage and go to the next higher standard voltage rating.

Enjoy your build.