My amplifier

Discussion in 'DIY' started by Alan0354, Mar 29, 2017.

  1. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    I want to explain a little on the OPS stage in my amp. I use the Folded 3EF as shown in the drawing where the top 3EF is PNP-NPN-NPN, bottom is NPN-PNP-PNP.

    Fold 3EF.jpg

    The advantages are:

    1) Output can swing closer to rail. The OPS can swing to 1Vbe from both rails instead of like 4Vbe from both rails. That's 3Vbe or at least 1.7V gain in voltage swing on each side OR 3.4V total swing advantage. For Class A or high bias Class AB that dissipate a lot of power, you need to use lower rail voltages. This can make quite a bit of difference if you can swing 1.7 closer to the rail.

    2) As circled in green, the pairs are screwed together on a separate heatsink, the variation of Vbe of the two transistor due to temperature change literally cancel out. The bias spreader only has to compensate for the Vbe of the output transistors. This is a big advantage over the convention 3EF where it's NPN-NPN-NPN on top and PNP-PNP-PNP at the bottom where you have to compensate for all 6 transistors. I have two amp with this design and the bias vary about 15% from stone cold to burning hot( driving load for period of time). In normal operation, the bias drift is much lower. So this amp does not need a lot of warmup.

    3) Folded 3EF has natural output short circuit current limiting. I design with tail current of the first transistor( pre-driver) of only 10mA, if you calculate assume hfe of the follower two stages are 50 each transistor. The total current gain of the driver and output transistor are 50 X 50=2500. So the current limit of the output DC current is 10mA X2500=25A. It is high current, but not as high if you have the 9 output pair of regular 3EF that can pump like 100A!!!!


    If you look at the diagram and follow the voltage starting with output equals 0V. The Vbe of the big output transistors and driver transistors are typically 0.55V only. If you follow the voltage written in red, you'll see the voltage across the bias spreader is 2.2V-Vbe1 and Vbe2. Small transistors normally has higher Vbe like 0.7V, so the bias spreader voltage is 2.2-2X0.7= 0.8V.

    Bias spreader needs more voltage to establish a stable voltage, 0.8V is not enough from my experiment. That's the reason I need to choose the pre-drivers with low Vbe. As I said, higher current and bigger transistor has lower Vbe, so I have to use big TO-220 transistors for pre-drivers.

    that's why it's a lot more complicated than just find another transistor. It affect the thermal stability of the amp. I have enough for my life time of hobby of these transistors, but I did not figure into anyone want to build my amp!!!
     
    Last edited: Jul 14, 2017
  2. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Thanks

    It is because of my amp, I try to build as high quality as I can. If you look at the pictures of my first amp in this thread, almost all wires are screwed onto either terminal block or to connector. Only the power wires on the terminal blocks, the output screw directly onto the output banana connectors. All the low power wires and signal are connected by Molex connectors. I think there is only two or three wires soldered. You can take the whole amp amp apart with a screw drive and a wrench. This is my pride and joy, has to be top quality on everything, no compromise.

    I use a lot of 12 gauge wires, no skimping on money on this. You should look into buying some big wires too. I hate that inside of a nice amp like my Nakamichi are tiny little wires running for power and even output.
     
    kjello likes this.
  3. kjello

    kjello AK Subscriber Subscriber

    Messages:
    299
    Location:
    Trondheim, Norway
    I wish I could understand half of this
    I need my book now
     
  4. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Yes, it is described in Cordell's book called Diamond Driver in page 191. But mine is different in the sense the collectors are going to the opposite rail instead of what is described in the book. I experimented the Diamond in my first try pcb, I found it is not very stable because the collector feed signal back and cause oscillation. So I change to what I have.

    There goes to show......you have to trust.........BUT.......VERIFY!!!! Never take anything at face value.

    There is nothing ground breaking in my design, the two books concentrate on lowering THD, so I designed my amp to be low THD. I studied through Cordell's book step by step how he lower THD, I used almost everything described in the book EXCEPT the cascode VAS described in page 68. I go for high bias that makes the amp very hot, I need to lower the rail voltage to lower the heat. Lower voltage lower the max power I can get due to limit voltage swing at the output. So I need to get the output swing to as close to the rail as possible. Cascode VAS eat at least 2V from the rail and it's not acceptable to me. And being 3EF, improvement from cascode VAS is questionable.

    Even most ( if not all) of the things in the book is not ground breaking as I said before that power amp is just a basic discrete IC opamp that had been well research and written in books on IC design.
     
    kjello likes this.
  5. Trojan4Life

    Trojan4Life Active Member

    Messages:
    377
    I'm trying to name the amp. How is THE ALAN0354?

    That is some awesome work you've done! Please send me info on cost to my email address.

    Meanwhile- I'll continue using the Marantz 1030 I have. It survived four years in college with my kid.
     
  6. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Ha ha, now you make me nervous!!! What if you don't like the sound, then you start throwing stone on my name!!!!

    I don't even name my amp!!!!

    I just received the shipment from Mouser, I am in business again. It was stalled for almost a week lacking parts.
     
  7. Trojan4Life

    Trojan4Life Active Member

    Messages:
    377
    That won't happen- I'm quite sure I'll like the sound, my old stuff is still sounding good. My musician son likes the sound of the old gear, that's a real compliment. When you compare to good live music, it's difficult to match. I'll send a photo of the faceplate when I'm done. Have a good weekend!
     
  8. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Ha ha, that make one of us.

    Make sure you have very good speaker cable.....or make cheap cable as shown in my other "cable and THD" thread using 5 or more pairs of 16 gauge speaker cables twisted together. Inferior cable that cause distortion negate all the effort to build a very low THD amp. It's about less than $30 to make my cable. AND just make it long enough to reach the speaker only.
     
  9. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Hi Kjello

    This is the chassis I used in my first amp, I actually bought and received the second one. I just decided to use my smaller chassis as shown in the picture in page 1 of this thread ( the black one).http://www.ebay.com/itm/BIG-Aluminum-A50-power-amplifier-chassis-suitable-for-KRELL-50w-class-A-circuit/291954201832?ssPageName=STRK:MEBIDX:IT&_trksid=p2055119.m1438.l2649
    I bargain with him and I bought the chassis for $325 total including shipping and everything. I think it's the same size as what you picked. The difference is mine has handles ( the two black thing on the front panel) which turn out to be very important. It's going to be heavy after you complete the build, it's big and clumsy, the two handles really become handy. Most importantly, this one has the hole for the power switch ( square hole), it's a lot harder to make that square hole.

    One thing, the second chassis I bought does not fit as good, I have to open a few holes a little. But I make the seller refund me $20. So I really paid only $305.

    Both the one you chose and mine are deceivingly big, the picture do no justice. You'll see when you get it. That's why the handle is important. I don't have the big wing span, I found it hard to pick it up without the handle. I only have a 625VA transformer, the 800VA is going to be a few pounds heavier. My first amp is about 52lbs, expect the one you build to be 60lbs.

    That's the reason I decide to use the black chassishttp://www.ebay.com/itm/Douk-Audio-...e=STRK:MEBIDX:IT&_trksid=p2055119.m1438.l2649 and use a 25V-25V 600VA transformer. It's a lower power amp, but the chassis is much smaller, easier to carry around. I want to take it out to stores one day and use the excuse of testing some good speakers with my own amp to listen and compare with other amps. Can't do that with my first amp as it's way too clumsy to haul it around.
     
    Last edited: Jul 16, 2017
  10. kjello

    kjello AK Subscriber Subscriber

    Messages:
    299
    Location:
    Trondheim, Norway
    Ok, cool
    I will make the handles and holes myself
    I can get 2 of the black chassis I linked for around 580 incl shipping
    And I like how they look. The empty chassis is 33 lbs
    They are the same size as my other amps
    *
    Im a bit concerned about damage during transport, Ive heard that is an issue with packing/transport from far east
     
    Last edited: Jul 16, 2017
  11. kjello

    kjello AK Subscriber Subscriber

    Messages:
    299
    Location:
    Trondheim, Norway
    Are you planning RCA input or XLR ?
     
  12. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Those black chassis I use is lower power, I have to use lower voltage transformer. The output is going to be around 64W@8ohm and 128W@4ohm. With the big chassis, you can use 30V-30V 800VA transformer to get 90W@8ohm or 180W@4ohm.

    I cannot split the amp into two chassis so each is a monoblock. Each channel has to be on one heatsink. Only way to do monoblock in smaller chassis is to do Bridge amp, I have not experiment with it yet.

    I am using simple RCA input. XLR is differential input for better hum reducing. My amp is very good already, XLR is just more complicated and might raise THD due to extra circuit.
     
  13. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    I totally forgot, what are the voltages of the rails? I designed my amp to be lower voltage rails and high bias. The driver transistors cannot take more than +/-50V rail. I prefer +/-40V max as power dissipation will be too high. For +/-40V, the power dissipation will be about 160W total already, it will get hot.
     
  14. Trojan4Life

    Trojan4Life Active Member

    Messages:
    377
    I'm buying a different transformer. I've got a couple from old gear I've worn out, such as a transformer from an old CEC DVD player which blew a component on the board and a new board would have cost as much as the DVD player cost when I bought it. Hardly play DVDs nowadays.
     
  15. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    I bought this already from Amtek. This is a 800VA 30V-30V. Rectified DC is 42V according to my first amp.http://www.antekinc.com/an-8430-800va-30v-transformer/

    For lower voltage, I bought this for my new amp 600VA 25V-25V. Rectified to 33V. http://www.antekinc.com/an-6225/

    I run 0.9A bias current, for 42V, the dissipation is 2X42VX0.9A=76W/side or 152W total. For 33V, dissipation is 60W/side or 120W total.

    Calculate how much power you can dissipate.
     
  16. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Actually, if you buy the torroid transformer, it will fit under the First Filter pcb. I designed the pcb to be right over the torroid transformer.

    ADD IPS filter 1.JPG

    This is my new amp. The First Filter board is the old version board shown in my first amp in post #22 in the second page of this thread. But the assembly is exactly the same as in this picture.

    I am at the process of writing the step by step how to build this and the wiring to get the one point ground correct with the design. The cap use depend on how tall the chassis is. Kjello plans to buy a chassis the same size as my first amp, he can use 10,000uF 63VDC and have more filter caps than me. My smaller chassis limits the height of this assembly to 5" tall, so I have to use combination of 4,700uF 100VDC in the First Filter and OPS Filter board, then I use 10,000uF for the IPS board. If your chassis has less than 5" tall, you might have to use shorter caps. The diameter and lead spacing of the caps are very common, you have a lot to choose from. The pin spacing is 10mm, you can find a lot of caps with this footprint to fit the pcb.
     
    kjello likes this.
  17. kjello

    kjello AK Subscriber Subscriber

    Messages:
    299
    Location:
    Trondheim, Norway
    Do you think I should order the chassis and transformer now? I expect 4-5 weeks of delivery time for the chassis.
     
  18. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    Give me few more days, The 2SC4793 is arriving on 7/26. I really want to power up one channel before you spend any money. You have more faith on me than......ME!!! I want to make sure it's working.

    As soon as I get the transistors, it should be fast to complete the whole channel, run THD, stability and all.
     
    kjello likes this.
  19. kjello

    kjello AK Subscriber Subscriber

    Messages:
    299
    Location:
    Trondheim, Norway
    Take your time, want it good, not fast :thumbsup:
     
  20. Alan0354

    Alan0354 Super Member

    Messages:
    2,087
    Location:
    Bay Area
    It's an honor you two want to build my amp, I am going an extra mile to make sure and also make it easy for you to build. While I am waiting for the transistors, I am taking pictures and writing the construction procedure so you can step by step following what I do. For you Kjello, you are buying the chassis similar to mine, you can literally implement the whole thing according to the procedure.
     
    kjello likes this.

Share This Page