Maplin MosFET Amplifier Ga28f construction thread

Bespoke Constructor
Joined 2006
Paid Member
Hi

Please find a few pics of my new Maplin 100w Mosfet amplifier construction thread .. All parts are new old stock with original transistors, but all capacitors are new, just a little bit of nostalgia from the past !!

Both amps will be mono blocks ..

best regards Nick
 

Attachments

  • P6080039.jpg
    P6080039.jpg
    493.1 KB · Views: 2,867
  • P6080040.jpg
    P6080040.jpg
    403.1 KB · Views: 2,730
  • P6080041.jpg
    P6080041.jpg
    400.8 KB · Views: 2,639
  • P6080042.jpg
    P6080042.jpg
    365.3 KB · Views: 2,630
  • P6080043.jpg
    P6080043.jpg
    440.9 KB · Views: 2,505
  • P6080044.jpg
    P6080044.jpg
    357.9 KB · Views: 997
  • P6080045.jpg
    P6080045.jpg
    456.7 KB · Views: 801
  • P6080046.jpg
    P6080046.jpg
    313 KB · Views: 841
  • P6080047.jpg
    P6080047.jpg
    309.3 KB · Views: 793
  • P6080048.jpg
    P6080048.jpg
    471.3 KB · Views: 825
Bespoke Constructor
Joined 2006
Paid Member
Bespoke Constructor
Joined 2006
Paid Member
Hi

I am having some new PCB's being made, 2 oz bare copper Fr4 boards .. I generated my own gerber files to update the original layout.

If anyone wants any let me know .. as the examples on ebay are over priced, I can make them very much cheaper with modern components .. however I do have a stock pile of the original transistors ... luckily for me

regards

Nick
 
Hi

I am having some new PCB's being made, 2 oz bare copper Fr4 boards .. I generated my own gerber files to update the original layout.

If anyone wants any let me know .. as the examples on ebay are over priced, I can make them very much cheaper with modern components .. however I do have a stock pile of the original transistors ... luckily for me

regards

Nick

Maplin amps on ebay are as rare as hens teeth.
I dropped out at £75 for a 225WRMS Maplin amp.

However, I picked up a pair for peanuts that were faulty a few months later.
 
Bespoke Constructor
Joined 2006
Paid Member
Hi

Yes that's why I have made my own full replicas 100% like the original with 2oz copper tracks. I have gerber files so I can now get them made any where that I like. My boards are also fully drilled ... I only needed one unmade unit so I could do the gerber conversion.

I purchased a bulk NOS lot of the original transistors ... I will try and put some more pics up later today of the build

regards

Nick
 
Disabled Account
Joined 2014
Just my input on this thread.

I bought this amp some time ago at a boot sale for £15.
I have not tried it out yet !!!!

It has 4 of the Maplin modules with the Maplin bridging / output protection unit in a dual mono configuration.


Kevin
 

Attachments

  • IMG_0295.JPG
    IMG_0295.JPG
    283.2 KB · Views: 1,406
Very nice amp to rival 90% of designs sold retail . The bias pot is a bit too big . 470 R is usually better . Exicon MOSFET's used later on . 10N/P20 I think now .

The MOSFET gate resistors can be increased to 220R . Use standard metal film as the small inductance helps . The gates oscillate at about 5 MHz . Often 220R stops it ever showing anywhere else . The output stage will accept at least two extra MOS FET's if difficult load . Make a longer bracket and add side by side . 220R to each gate . 2 x 1 x 1/4 stock aluminum works well.

100 mA is often said to be ideal bias ( brake power line and deduct 14 mA . ie 114 mA ) . Less than that will work fine, will sound brighter . Chooses the sound you like best . Unlike bipolar transistors there isn't the more perfect point . 100 mA is the theoretical tipping point of these FET's . That is when the famed Ron is optimum . Equally setting it lower gives slightly less Ron and helps drive . It is no big deal how it is set as long as not too hot . If you can not briefly touch the MOS FET can it is too hot and may require a larger heat sink ( spit on your finger , sizzle is not a good sign ) . 0.5 degree per watt per PCB would be ideal . Less usually is OK . Do finger test when running music . Again you should just about be able to touch it ( 80 C I suspect ) . The heat sink should never be above 60 C if possible . The amps typically are about 50 % efficient all told . Techo music is 30% crest factor ( Hypex say 10% for nice music ) . So a 4R 50 watts of heat . 25 C + 30 C ambient = 55 C which is OK . For 8 R 1 deg per watt per PCB seems possible .

The 47K resistor to the input PNP transistors is said to be the worst feature of this amp . Not true . Slewing is symmetrical which is rare and good .

The 220 uF decoupling caps can be enlarged and made low Z ( 470 uF 63V ) . The 47 uF feedback arm cap can be made 63V non polar . Input cap 10 uF film type . You can take your PSU ground to between these 220 uF caps just before the output devices , it is marginally better .

27pf to 2SD756 can be NPO high grades ( most are ) .

Gain can be increased to 100 using 330R and 150 uF . This will eliminate the need for a preamp . 470R will often be enough as the amp is powerful . Be careful of reducing gain . The most I tried was 15 ( 2K2 ) .

The industrial FET's can be used ( IRFP etc ) . They require thermally compensated biasing like bipolar amps according to most who have done it . Purists reject this idea as they say it defeats the who idea of the Hitachi design . Dave Goodman seems to have used 100 % the Hitachi , wise man .

Make no mistake this amp will eat a Naim NAP 250 for breakfast if having a top notch PSU . It will disgrace a Quad 405 . It will please valve amp lovers . Goldmund was a similar design . HH made a 1200 watt version . It's distortion is very low especially at high frequencies . Beware fake transistors if cloning it . They will ruin it . 2SA970 2N55511/5401 might work at a pinch . HH used MPSA 42/92 which seem oddly different . Nice devices so maybe OK . The 2SD756 2SB716 are special devices of extremely high gain ( 400 on my meter ) , also low Cob . The current mirror using 1N4007 is fine as is ( excellent as proved by 3K9 balance ) . The 2 x 3K9 on the input pair collectors should balance at about 2% . 1.98 and 1.96 V often being typical ( 500 mA ) . The 100R second pair tail at 1.4 V typical ( 2 - 1.4 = 0.6 V 2SD756 bias voltage ) . I think I calculated slew rate to be 35 V/uS which is OK , it is double what it would be if a standard single VAS ( 2SD's ) . 47K is a useful input resistance if using a 10 K pot to feed it . Can be made 33K to minutely improve DC offset ( about 20 mV which is fine ) .

Was also called LP 56 I think . The GA 28 was a slightly cheapened version that was 100% as good or better as used Exicon ( made in Scotland ) .
 
Last edited:
Very nice amp to rival 90% of designs sold retail . The bias pot is a bit too big . 470 R is usually better . Exicon MOSFET's used later on . 10N/P20 I think now .

The MOSFET gate resistors can be increased to 220R . Use standard metal film as the small inductance helps . The gates oscillate at about 5 MHz . Often 220R stops it ever showing anywhere else . The output stage will accept at least two extra MOS FET's if difficult load . Make a longer bracket and add side by side . 220R to each gate . 2 x 1 x 1/4 stock aluminum works well.

100 mA is often said to be ideal bias ( brake power line and deduct 14 mA . ie 114 mA ) . Less than that will work fine, will sound brighter . Chooses the sound you like best . Unlike bipolar transistors there isn't the more perfect point .

The 47K resistor to the input PNP transistors is said to be the worst feature of this amp . Not true . Slewing is symmetrical which is rare and good .

The 220 uF decoupling caps can be enlarged and made low Z ( 470 uF 63V ) . The 47 uF feedback arm cap can be made 63V non polar . Input cap 10 uF film type . You can take your PSU ground to between these caps , marginally better .

27pf to 2SD756 can be NPO high grades ( most are ) .

Gain can be increased to 100 using 330R and 150 uF . This will eliminate the need for a preamp . 470R will often be enough as the amp is powerful . Be careful of reducing gain . The most I tried was 15 ( 2K2 ) .

The industrial FET's can be used ( IRFP etc ) . They require thermally compensated biasing like bipolar amps according to most who have done it .

Make no mistake this amp will eat a Naim NAP 250 for breakfast if having a top notch PSU . It will disgrace a Quad 405 . It will please valve amp lovers . Goldmund was a similar design . HH made a 1200 watt version . It's distortion is very low especially at high frequencies . Beware fake transistors if cloning it . They will ruin it . 2SA970 2N55511/5401 might work at a pinch . HH used MPSA 42/92 which seem oddly different . Nice devices so maybe OK . The 2SD756 2SB716 are special devices of extremely high gain ( 400 on my meter ) , also low Cob . The current mirror using 1N4007 is fine as is ( excellent as proved by 3K9 balance ) . The 2 x 3K9 on the input pair collectors should balance at about 2% . 1.98 and 1.96 V often being typical ( 500 mA ) . The 100R second pair tail at 1.4 V typical ( 2 - 1.4 = 0.6 V 2SD756 bias voltage ) . I think I calculated slew rate to be 35 V/uS which is OK , it is double what it would be if a standard single VAS ( 2SD's ) . 47K is a useful input resitance if using a 10 K pot to feed it . Can be made 33K to minutely improve DC offset ( about 20 mV which is fine ) .

The module should really have 220uf and 39R decoupling from the power supply to get rid of hum. Hum has been the complaint of a few of these amp owners.

I am currently waiting for pcb's for irfp240/9240 version.
It has a proper Vbe multiplier bias. I also added a DC offset adjustment.
I added the 390r and 220uf to each rail.
 
Maplin said use 100 R + 100 nF PSU ground to chassis to avoid hum loop . I have had - 100 dB hum . It seems to be a matter of thinking out the PSU . Then ground to the decouplers works . Never shift the Zobel ( 4R7 + 100 nF ) . Smoke comes soon after ( D Self idea that is a NO No ) .

A PSU for thew driver stage would be good .
 
ESP Audio says the same . There is about 3 mA at zero bias . Up to 10 kHz Rod Elliot got 10 kHz distortion free . Here are some notes I made years ago . This simple current source replaces R3 . It makes people who write at DIY Audio happy to see it . My own tests say R3 is OK . The other transistor is just how a current mirror usually is made up .

PdTBDyd.jpg
 
Two Nigel's hey? Nigel says something very valuable. The special MOS FET for audio might replace germanium in a Leak amp . The germanium 0.5 V bias for two devices will be enough to drop them into the old circuit without modification ( remove drivers ) . I dare say with a little care using a scope it will be a marriage made in heaven . Exicon seem to need 1.2 V typical to do 100 mA . That means better than any silicon device even when complimentary feedback pairs ( 1.3 V ) . Dalingtons 2.5 V . I could right a page on this and not say much more . The industrial FET's might need 4 V .

The function of R9 12 K and the 6n8 poly cap is to balance the VAS's at HF . It mimics the output devices . If the voltage is changed R9 needs changing . 6.5 mA 12 K = 78 V if +/- 52 V rails . You will loose the balance in the 3K9's if wrong . Tweak until about like here . Approximately reduce the resistor exactly as the rails . Thus +/- 40 V is 9K1 . R3 is 1K / V . It would need 39K for 40 V .
 
Last edited: