diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Pass Labs (http://www.diyaudio.com/forums/pass-labs/)
-   -   Balanced F5 in a Small Footprint (http://www.diyaudio.com/forums/pass-labs/213152-balanced-f5-small-footprint.html)

lhquam 23rd May 2012 04:14 AM

Balanced F5 in a Small Footprint
 
1 Attachment(s)
I present the Balanced F5 in a Small Footprint.

I have finally completed testing of my balanced F5 PCB which is optimized for use with a relatively small 5" x 7" heatsink. This effort with inspired by EUVL's F5X project http://www.diyaudio.com/forums/pass-...ml#post1538826 http://www.diyaudio.com/forums/pass-...?highlight=f5x, by my previous experience with the Cviller F5 V1.2 boards on a 5" x 7" heat tunnel: http://www.diyaudio.com/forums/pass-...ml#post2928566 and by my experience using a heat tunnel http://www.diyaudio.com/forums/pass-...ml#post2453420.

I envision five different configurations for using this PCB:

A. Heat tunnel, stereo balanced F5, 17 volt rails, 40-50 W/ch class A. Shown in image below.

B. Heat tunnel, four F5s, 24 volt rails, 25 watts/ch class A.

C. Heat tunnel, stereo F5 Turbo V1, 32 volt rails, 40-50 W/ch class A. PCB partially populated.

D. Heat tunnel, 1 channel balanced F5, 32 volt rails, 100 watts class A, 200(?) watts class AB. PCB partially populated.

E. Small heatsink, headphone amplifier, two F5s, ?? volt rails, ?? watts/ch class A.

I am primarily interested in the stereo balanced F5 configuration A and will concentrate on it in this thread. I couldn't decide what MOSFETs to use for the balanced amplifier of configuration A, so I populated one PCB using FQA28N15/FQA36P15 MOSFETs and another using FQP19N20C/FQP12P20 MOSFETs. All of these MOSFETs are still available from DigiKey and Mouser. I intend to test another board using Toshiba 2SK1530/2SJ201 MOSFETs which are made of virtual unobtainium.

Following posts will show PCB design, packaging and performance results for two configurations:

. FQA28N15/FQA36P15 MOSFETs with Caddock source and feedback resistors mounted to the heatsink.

. FQP19N20C/FQP12P20 MOSFETs clamped to the heatsink and onboard wirewound source and feedback resistors.

lhquam 23rd May 2012 04:17 AM

Chassis design
 
2 Attachment(s)
The following images show amplifier without cover. The dimensions are 11" W x 9" H x 13" D. An alternate chassis design is presented in a later post.

lhquam 23rd May 2012 04:21 AM

PCB V1.0
 
4 Attachment(s)
The V1.0 PCB contains two F5 amplifiers with the circuit topology shown here and on page 2 of Nelson's F5 Turbo article http://www.firstwatt.com/pdf/art_f5_turbo.pdf. Version V1.0 was designed and fabricated using ExpressPCB ExpressPCB - Free PCB layout software - Low cost circuit boards - Top quality PCB manufacturing 2.5" x 3.8" 2-layer MiniBoards.

The 2nd image shows the PCB design with outboard MOSFETS and source and feedback resistors in relation to the heatsink (the surrounding yellow rectangle), with the airflow from bottom-to-top. The MOSFET placement was carefully chosen for heat distribution. The design is for 100-120 watts dissipation per heatsink. I borrow heavily from EUVL's elegant PCB layout and its use of standoffs for power and output connections. Two 2-pin headers provide input connections. Two 6-pin headers provide "diagnostic" connections for bias adjustment.

Version V1.0 of the PCB can be used to implement an F5X channel, a balanced F5 channel, or two independent F5 channels. There are 2 major, permanent jumpers for the X connections. Without any other jumpers, the board provides an F5X. Two jumpers provide a floating balanced togolopy, (the X's are connected). Two more jumpers activate Nelson's P3 adjustment. And finally, two more jumpers ground everything for initial adjustment or for use as 2 independent unbalanced F5 amplifiers. Originally, the jumpers were intended to be made using .1 inch shunts, but I abandoned that approach because of lack of good contact pressure and uncertainty about the the contact resistance of the shunts.

lhquam 23rd May 2012 04:25 AM

PCB with FQA28N15/FQA36P15 MOSFETS
 
3 Attachment(s)
The first PCB was built using FQA28N15/FQA36P15 MOSFETs, .25 ohm Caddock source resistors, and 50 ohm Caddock feedback resistors as shown in the first image. Because of the limited power dissipation requirements, I mounted the Caddocks vertical to the PCB. I later found that this is a bad idea because the leads on the Caddocks are fragile and break after a few flexes. I "repaired" the Caddocks and mounted them to the heatsink using clamps as shown in the 2nd and 3rd images.

lhquam 23rd May 2012 04:27 AM

PCB with FQP19N20C/FQP12P20 MOSFETs
 
1 Attachment(s)
The second PCB was populated with FQP19N20C/FQP12P20 MOSFETs, mounted using clamps, and .25 ohm wirewound source and 50 ohm wirewound feedback resistors. Caddocks could have been used, but I wanted to test this alternate resistor configuration.

lhquam 23rd May 2012 04:29 AM

Bias adjustment
 
2 Attachment(s)
To simplify the adjustment of bias current and output offset current, I have provided two 6-pin headers
on the PCB for attaching a "diagnostic cable". This cable is wired to a 6-position, 2-pole rotary switch, connected to binding posts for a multimeter. The rotary switch positions select the following:

1. Vs1+ voltage across P-channel source resistor of amplifier 1
2. Vs1- voltage across N-channel source resistor of amplifier 1
3. Out1 output voltage of amplifier 1
4. Out2 output voltage of amplifier 2
5. Vs2+ voltage across P-channel source resistor of amplifier 2
6. Vs2- voltage across N-channel source resistor of amplifier 2

lhquam 23rd May 2012 04:45 AM

Distortion Measurements
 
4 Attachment(s)
THD measurements were made using SillanumSoft Visual Analyzer 2011 XE Visual Analyser 2011 XE
and an M-Audio Audiophile 192 sound card. The amplifiers were measured in grounded balanced mode with an 8 ohm resistive load. Four sets of graphs, each containing curves at 1.0A, 1.5A, and 2.0A bias currentsare shown:

1. FQA28N15/FQA36P15 THD vs Power at 1 kHz.
2. FQP19N20C/FQP12P20 THD vs Power at 1 kHz.
3. FQA28N15/FQA36P15 THD vs Frequency at 2 watts output.
4. FQP19N20C/FQP12P20 THD vs Frequency at 2 watts output.

At low power levels, the FQP19N20C/FQP12P20 MOSFETs provide lower THD levels than the FQA28N15/FQA36P15 MOSFETs. At power levels above 40 watts amplifier clips, primarily due to the lower transconductance of these MOSFETs. At 40 watts, the THD is similar to that of the F5 Turbo at 40 watts, but much better at low power levels.

With the FQA28N15/FQA36P15 MOSFETs the amplifier produces 50-55 watts with THD around .2%.
This is somewhat higher than Nelson shows for the F5 Turbo, probably due to lower MOSFET transconductance.

Conclusion:

For maximum power, my choice is the FQA28N15/FQA36P15 MOSFETs at around 1.5A bias.

For lowest THD at 1-2 watts, my choice is the FQP19N20C/FQP12P20 MOSFETs at around 2.0A bias.

lhquam 23rd May 2012 04:46 AM

Power Supply
 
1 Attachment(s)
An Antek AN-5415 toroid transformer with four 15 volt secondary windings, is used to power two +16V and two -16V supplies. Each supply uses 1/2 of a 35A 200V Vishay diode bridge, four .022 Farad 35 volt electrolytic capacitors, and a .5 ohm Caddock resistor, forming a CRC filter.

With the diode and CRC voltage drops, the nominal rail voltages are +/-16.9V with 2.0A bias, and 17.9V with 1.0 bias.

I measured A-weighted noise across the balanced outputs. Both channels gave essentially identical results:

With shorted inputs:
A-weight noise = 70 uv. 60 Hz harmonics less than 20 uv.

With the amplifier powered off:
A-weight noise = 70 uv. 60 Hz harmonics unmeasurable - near noise floor of sound card.

lhquam 23rd May 2012 04:47 AM

PCB V1.1
 
1 Attachment(s)
Version V1.1 of the PCB (not yet fabricated) contains three major changes to V1.0:

1. Drop support for the F5X option, supporting only the floating balanced F5, the grounded balanced F5, and the stereo (unbalanced) F5 options. With this change, only one "jumper screw" is required for grounding the floating node. A large (#4 screw size) plated-thru hole connects the floating node (connected X's) on the bottom side to ground on the top side. In order to "float" the node, the plating is removed around the top of plated-thru hole using a slightly larger drill bit. Place a #4 screw thru the hole, soldering its head to the floating node on the bottom side. The node is re-grounded screwing a #4 nut onto the screw, tightening to make contact with ground on the top side of the board.

2. Combine the two 2-pin input headers into a single 4-pin input header.

2. Increase the diameters of many plated thru holes, making component replacement easier.

lhquam 23rd May 2012 04:48 AM

Alternate Enclosure
 
1 Attachment(s)
After fabricating the chassis shown in post #1 decided to try different aspect ratio design having
dimensions 11" W x 6" H x 15" D. Drawings are shown here:


All times are GMT. The time now is 10:59 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2