Hybrid Class D System
started on October 3 ^st , 2024

INTRODUCTION

Many years ago I conducted a series of tests on class D amplifiers, four products were listened to in 3 different condictions and with different persons.
Even if the sound of each of these products was different from the others, none of them could come close to the sound result obtainable with a good quality traditional class AB or class A amplifier.
The technology of class D amplifiers has evolved and now I hear positive opinions from my audiophile friends even in comparative listening sessions with traditional amplifiers.


FOSI FOX V3

There is an amplifier that is considered a small miracle, I am speaking about the Fosi Fox V3 with the chip TPA3255.
It is a very cheap product but Fosi Audio has used audio grade components and a good layout.

The opamp are DIP on socket so these are wappable and you can test different products to get your preferred sound signature.
This amplifier does not use the PFFB (post-filter feedback) that is present in the V3 Mono version but in this product there are more opamp on the signal path.
PFFB offers many benefits including lower output noise, improved THD+N performance, improved IMD performance, lower output impedance, frequency response less affected by load impedance, and suppression of nonlinearities of  the LC filter (I will test in the future).

The declared specification are:

• Max Power Output: 300Wx2 @4Ω
• SINAD: ≥88dB
• SNR: ≥110dB
• Terminating Impedance: 2-8Ω
• THD: <0.003%
• Power Amp: TI TPA3255
• Op-Amp: TI NE5532 x2
• Power Supply: 32V/5A, 48V/5A
• Power Input Range: DC 24V-48V
• Frequency Response Range:  20Hz-20kHz(±0.1dB)
• Input: RCA

The amplifier is composed by 3 parts:  input stage, a class-D power amplifier (TPA3255) and an external power supply.
The input stage is necessary to produce a balanced signal to drive the second stage.
The schematic of Fosy Fox V3 is very near to the evaluation board of TPA3255.





The only difference between the 2 schemes is represented by the capacitor in the middle to separate the 2 opamps but it is not essential because both points are almost at the same voltage.
One thing that an audio engineer sure doesn't like is the opamp powered by a single power supply (12V), the non-inverting input in this case is connected to a half voltage reference (6V).
This input stage is not the best way to create a differential signal with opamps but it is the only one that requires only a double opamp and keeps the input inpedance high so it does not require an additional opamp to act as a input buffer.
The interstage capacitor on Fosi Fox V3 are 8 x 10uF Elna RA3, this is a cheap audio grade cap. far from Elna SILMIC (in any case too soft in my opinion).
I prefer the sound of Nichicon Muse UFG but replace all this caps is not very easy.
The output filter is composed by Sumida Audio Inductors and WIMA Red Capacitors.
This amplifier come with an external 32V switching power supply but there is the option to use a 48V to increase the output power.
The most important and easy intervention concerns the replacement of the opamps.
I have tested the following models:

Company	model	CMRR (dB)	KsVR (dB)	SR (V/us)	Band (MHz)	Aol (dB)	cost ($)
Texas Instruments	NE5532	100	100	9	10		1-2$
Nisshinbo	Muse02	110	110	5	5.8		40$
SparkosLabs	SS3602			20/12	5	140	80$

CMRR: Common-mode rejection ratio
kSVR:   Supply-voltage rejection ratio (ΔVCC±/ΔVIO)
SR:       Slewrate at unity gain
Band:   Unity-gain bandwidth
Aol:      Open Loop Gain 

The listening tests were carried out with the greatest possible objectivity and using a very analogue digital source, my NAA DSD DSC with HQPlayer software set for a resample to 5.6MHz DSD.
The Muse02 sounds better than the original NE5532 included in the Fox V3 but only because it makes the sound softer but it cancels out many details and adds a tail to the bass.
The NE5532 is sometimes too aggressive and it can certainly be a problem especially with normal digital sources, in this case the Muse02 can help a lot.
The SS3602 is very different from the first chips and this is immediately noticeable with any song.
It does not sound aggressive but maintains all the details with a dry bass without tail, so it is definitely the final choice even if the cost is high.

The Fox v3 is sold with a 32v 5A switching power supply but it is possible to buy the more powerful 48V 5V version with which the amplifier sounds decidedly cleaner even at normal power.

The second listening session was done using a very simple tube pre amplifier composed of a ECC82/12AU7/5814 Jan Philips tube and a IRF820 mosfet (this pre-ampl is in develop phase but you can see also the my last Headphone amp).
The preamp has greatly increased the overall sound effect, the depth of the soundstage and the harmonics of the instruments and eliminates the aseptic effect typical of solid state amplifiers so I suggest it.
I suggest the use of vaccum tube pre-ampl. with low voltage gain, no feedback and a low output impedance.

So even if it seems incredible a Fosy Fox V3 with 48V power supply, a pair of SS3602 opamp and a good simple preampl. can give a sound comparable to one of my latest amplifiers.

I don't suggets the Fosi V3 mono becasue there is one more opamp on the signal path to create the useless balanced input.

The complete project of the tube pre-ampl used in this test will be published soon and a new pcb will be tested.

In this article there are no measurements on the class D amplifier because these are not useless in this system because the tube preamplifier will have much higher distortion than class D section.

In the future I would like to eliminate the opamps on the signal path and use a differential vaccum tube preampl like my DAC output stage (in develop phase).

The interconnect cables are NEOTECH NEMOI-3220 and SynchestraRef by Luminous Audio.

TUBE PRE-AMPL

• Voltage gain 14x  about 23dB

• Frequency response > 1Mhz

• Output impedance 5ohm

• Zero noise with Virtual battery power supply

• DC filament

• Soft start filament

• No feedback

• Switch-on relay anti-bump

• Use output capacitors 10uF min to drive 10Kohm load

R18,R19,R42,R43         220K 1/8W  1%  RN55D2203FB144
R59,R60,RR65,R66        100K 1/8W  1%  RN55D1003FB14
R31,R32,R61,R62          33K   2W  5%  PR02
R5,R6,R69,R70             1K 1/8W  1%  RN55C1001BB14
R3,R4,R33,R35            82K   1W  1%  PR01
R20,R21,R44,R49           2K 1/8W  1%  RN55D2001FB14
R12,R30                 470K 1/4W  1%
R13,R28                  10K   2W  5%
R14,R29                 150K   3W  5%
R2,R38                  1200 1/8W  1%  RN55C1201BB14
R45                      68K 1/4W
R48                     1800 1/4W
R46                      150 1/4W
R47                      100 multi-turn trimmerr
R71                     220K 1/4W
R72                       1K 1/4W
R74                     2200 1/4W
R73,R11,R34             4700 1/4W
R75                      47K 1/4W

C2,C4                   33uF 400V   UVY2G330MHD
C3,C5                  220uF  10V   UFG
C6                     100uF 400V   LGU2G101MELZ
C7                    4700uF 25V   
C1,C8,C10,C11          220uF 25V
CY5,CY6,CY7,CY8         10nF 440VAC  R474I210050A1K V057
U$5,U$8,U$9,U$11        12uF 400V MKP (min. 10uF)

U$3,U$6                IRF840
U$19,U$16,D6,D8        1N4007
Q5,Q6                  IRF820
T3                     2N2905
U$15                   LM393
B1                     2KBP01
D5,D10                 Zener 10V 0.5W
D1,D2,D3,D4            UF5407
T1                     BC517 (invert pin E with C)

RL1                    G6S-2 2-pole micro signal relay

Connections            RS 718-7987  MOUSER 571-1217056-1

5 item of Heat Sink TO220 Vertical 11 Degree C/W, 2.67mm Hole, 38.1x34.92x12.7mm Aavid 513102B02500G Mouser 532-513102B25

Here the files to produce the PCB

Here the system with temporary set of transformers, you can use a single transf. with 230V 0.2A and 15V 1A.

I have used a 5814 Jan Philips to get the best sound but you can use any 12AU7/ECC82/E82CC.

Here you can see the frequency response (> 1Mhz) and voltage gain (both channels) of this pre-ampl.