Premium Sweetie Amp: Technical Talk, Parts Options, Testing Results and Other Comments

[captainbeefheart]

3 hours ago, tube fanatic said:

Sorry Cap, but your arguments are only partially correct.  But, this is not the venue for such a technical discussion.  If I can find time during the week between Christmas and New Year, I will start a thread on DIY Audio and discuss it with you there, where it belongs (I assume you are still very active on there as well).  We can post a representative amplifier and analyze every factor from input to output and look at what happens everywhere.  Then, we can have the input of other EEs to really spice things up a bit!  

 

Maynard

 

Of course there are stability criterion for feedback. You can't just go adding it all willy-nilly.

 

Take three cascading stages all AC coupled. If each of the coupling capacitors produces a phase difference of 60° at the same frequency, the total phase shift adds up to 180°.  This amounts to phase reversal. If the feedback starts out as being negative for this particular frequency it will convert into positive feedback. The critical point is a loop gain AB = 1

 

With the Sweetie being only a two stage amplifier and combined with the negligible amount of feedback I added there were no real reasons to bring up stability.

 

I really don't see a need to bring this conversation to another forum. I'm sure there are plenty of guys in here interested in the topic and our conversation could be educational for them.

 

 

[tube fanatic]

At this point I can only say that this is Henry’s thread and he can decide how technical he wants it to be.  I believe that most participants are more interested in learning about equipment which complements their Klipsch speakers than knowing why.  If I am wrong, then start a thread on tube amplifier design and I will try to be involved as time permits.  
 

Forum members, please post your thoughts on whether you want to learn about tube amp design and Henry and the moderators can make the appropriate decision about continuing the discussions here.

 

Maynard

[henry4841]

1 minute ago, Travis In Austin said:

I would like to keep this thread clean and in accordance with the the poster’s (Henry’s) original intent - a simple tube amp that anyone can build as a beginning project at a reasonable cost. 
 

@henry4841 would you mind if I split the technical discussions that follow the completion of the Little Sweetie into a separate thread where you all can continue the deep dive? This would allow those coming into this original thread to come in, follow your build, ask build questions, and post their builds?

 

Travis

Sure, go ahead.

 

Personally I like the circuit as it was designed by Maynard. The last few days, almost a week now, I have played with some changes in the circuit just because. Nothing I tried improved the circuit to any significant degree. I think Maynard got it right the first time many years ago. I have not tried feedback being I see no reason to reduce the amount of 2nd harmonics. That is about the only distortion seen on Arta and I like seeing it in a SET amplifier. That is why I build SET amplifiers. All I can add is if and when someone builds this amplifier they can play with it all they want and publish their findings here or on another thread. I do not think I am going to play with this circuit anymore myself. I like what I hear. This build is about as cheap one can build a quality amplifier and I do not believe you will be disappointed. The original Sweetie project that is. No need to go to the extreme I did on this premium build. Yes the premium may sound somewhat better but not enough to get excited about. The original sounds plenty good enough. Just use good Digi-key or Mouser parts. 

[captainbeefheart]

I was only sharing my experience on building the Sweetie, well half a sweetie.

 

As I mentioned I used an inexpensive "universal" SE output transformer with the circuit and since my bandwidth wasn't great I decided since the circuit had enough gain to just add some feedback. It's such a small amount of feedback that it shouldn't cause any trouble around just two stages. I found that a little feedback improved things for my situation.

 

By no means is adding -7db of feedback going to drastically change the amplifier into something else. I did get lower distortion of course but there is still plenty of second harmonic distortion, it certainly didn't disappear. I was more concerned with bandwidth and output impedance which is why I added the feedback.

 

I am in no way trying to change the original design at all, nor am I saying it's a bad design. Someone else may also not use the best output transformers like me and find the feedback helps their situation. With cheapo output transformers take care not to use too much feedback unless you know what you are doing, the -7db with the 2.2k resistor is a safe amount and shouldn't cause troubles.

 

 

[captainbeefheart]

Advice to anyone building the Sweetie or any other zero feedback SET amplifier is if you want excellent performance and sound then I would think long and hard about the output transformers and make certain you use a very high quality one. Another member mentioned Lundahl and I second that, their amorphous C core output transformers are excellent.

 

If it's just a fun project and ultimate sound quality isn't necessary then use whatever transformers you want. It was really no surprise to me that the output transformer I used didn't give the best bandwidth. If you find your build is lacking in treble frequencies and you are not an ideologue against negative feedback then add the 2.2k resistor and see if there is an improvement. It won't ruin the SET sound contrary to belief. It's such a small amount of feedback, and around only two stages there shouldn't be need for compensation. Trust your ear and use whichever you like best.

 

https://www.lundahltransformers.com/wp-content/uploads/datasheets/1663.pdf

[henry4841]

11 hours ago, captainbeefheart said:

Advice to anyone building the Sweetie or any other zero feedback SET amplifier is if you want excellent performance and sound then I would think long and hard about the output transformers and make certain you use a very high quality one. Another member mentioned Lundahl and I second that, their amorphous C core output transformers are excellent.

 

https://www.lundahltransformers.com/wp-content/uploads/datasheets/1663.pdf

 But then you will defeat the purpose of the build. Those Lundahl transformers are expensive. I plan on doing some more test today to see when the high frequencies are down 3db using the Hammond 125Dse's like the ones used on this build. I really like the sound of the Hammond 125 series of transformers I have tried of late over the Edcor's. I have two Sweeties right now, one with Hammond OPT's and the other with Edcor to compare the sound with. The one with the Edcor's is a stereo version whereas the premium Sweetie is dual mono's on one chassis though. Should not make any difference in frequency range test though.  

 

One can always improve any amplifier to some degree by throwing expensive parts at it. The goal of the Sweetie was a really good sounding amplifier for the lowest cost using good quality parts. I honestly think the Sweetie as designed by Maynard fulfils that goal as a good inexpensive forum amplifier project. One thing I like is the lower voltage it operates on making it more appealing for a beginner to build. That and no adjustments are necessary if one follows the build guide. 

[tube fanatic]

14 minutes ago, henry4841 said:

 But then you will defeat the purpose of the build. Those Lundahl transformers are expensive. 

 

One can always improve any amplifier to some degree by throwing expensive parts at it. The goal of the Sweetie was a really good sounding amplifier for the lowest cost using good quality parts. I honestly think the Sweetie as designed by Maynard fulfils that goal as a good inexpensive forum amplifier project. One thing I like is the lower voltage it operates on making it more appealing for a beginner to build. That and no adjustments are necessary if one follows the build guide. 

The Hammond 125SE series transformers are capable of outstanding bandwidth and specs when used well below their rated power.  As Henry said, one can always spend more.  We have all seen the ad nauseam discussions about beeswax filled capacitors, etc.  George Anderson ran measurements on the Hammond 125CSE which clearly show just how excellent a budget unit can be if used appropriately.

 

http://tubelab.com/articles/component-testing/budget-output-transformers/

 

Maynard

[henry4841]

5 minutes ago, tube fanatic said:

The Hammond 125SE series transformers are capable of outstanding bandwidth and specs when used well below their rated power.  As Henry said, one can always spend more.  We have all seen the ad nauseam discussions about beeswax filled capacitors, etc.  George Anderson ran measurements on the Hammond 125CSE which clearly show just how excellent a budget unit can be if used appropriately.

 

http://tubelab.com/articles/component-testing/budget-output-transformers/

 

Maynard

I forgot George has already ran some test on Hammond 125 series transformers. The numbers speak for themselves. I talk to George on occasion and he is talking about taking his site down. I would hate to see that happen but I can understand why. below is what he found on the Hammond 125Cse's at his website. Quote,

 

 

 Hammond 125CSE

These have been the best choice in low cost transformers. Best sound is achieved at 70mA. This transformer makes about 1 watt more above 100Hz than the Electra-Print. Specs at 1 watt: frequency response is 11.7Hz to 42.9KHz with 50mA, and 10.4Hz to 44.2KHz with 70 mA. Distortion is .65% with 70mA (mostly 2nd harmonic). Distortion is 2.6% at 6.0 watts. Low frequency effects at 5 watts: Visible saturation can be seen on the scope for any frequency below 77 Hz at 50 mA, 64Hz at 80mA. clipping was observed at any frequency below 130Hz with 80mA, but none was noted at 50 mA. Low frequency distortion tests at 1 watt: 100Hz lowest distortion was 1.01% with 63 Ma. At 36 HZ .66% occurs with 58mA.

 

[captainbeefheart]

Hammond is decent, they are $63 each so that's $126 for a stereo unit. That's not including shipping.

 

I was using a cheapo transformer which costs about $20 each, and I do not recommend.

 

George was measuring with a 300b tube at 70mA in a completely different circuit. The 300b has about half the rp as a 6Y6 if I remember correctly.

 

Do we have any Sweetie measurements with Hammond or Edcor?

[henry4841]

I ran some test using Arta this morning at 1/4 watt output and I am seeing approximately 2db down on one channel and 2 1/2db down on the other at 18khz with 1khz as a reference. This is on the Premium Sweetie build. I feel like most that would even consider this amplifier are like myself now limited on just how high a frequency they can hear. I did not notice an appreciable loss at 10k worth mentioning. Excellent numbers for frequency response in my world on a tube build. 

[captainbeefheart]

Thank you Henry for actual measurements.

 

Strange.....that is eerily similar to what I was getting. I was -3db down around 18kHz and assumed it was my OPT.

 

I started thinking what else would cause a roll off and it could very well be the 6Y6 input capacitance and the high output impedance of the 6SJ7 Pentode which the load resistance 100k dominates. Rough calculations of the RC low pass filter caused by driver tube and 6Y6 input capacitance and the -3db point is right around where we are measuring. 

 

Interesting, the high frequency roll off most likely isn't the output transformer but the actual circuit. 

 

Power triodes have very high input capacitance compared to pentodes and they require low impedance drivers in order not to get a roll off in the treble frequencies. 

[henry4841]

I do not consider the roll off at such a high frequency to be a problem other than for those that like to chase numbers. I consider 2db down at 18Khz to be very respectable. I can not hear that high anyways nor can most over 50. 

[captainbeefheart]

48 minutes ago, henry4841 said:

I do not consider the roll off at such a high frequency to be a problem other than for those that like to chase numbers. I consider 2db down at 18Khz to be very respectable. I can not hear that high anyways nor can most over 50. 

 

I wouldn't say it's just to "chase numbers",  harmonics from instruments such as cymbals, piccolo, and violin all extend up to 20kHz. The overblow/breath/air from female and even male vocals extend up that high. High fidelity is to recreate ALL the detail in the recording that anyone can hear.

 

I have excellent hearing for an old guy and I noticed it immediately. Which is why I added the feedback to increase the bandwidth.

 

The designer himself said that people are sensitive to phase shift and with -3db at 18kHz there will be significant phase shift in the audible spectrum.

 

Did you happen to measure the low frequency bandwidth? With the 300b in George's amp it was -3db at 11.7Hz and it's starting to roll off at 30Hz. The 300b has lower plate impedance so I figure the 6Y6 won't have the same roll off characteristic. My guess is it will be around -3db at 20Hz and start it's roll off much higher. Even with my cheapo OPT the bass on my full range mono setup wasn't bad, I only noticed how dull the recordings sounded which led me to look into the high frequency domain which showed I was losing the high frequency detail from the recording.

 

 

[henry4841]

41 minutes ago, captainbeefheart said:

 

 

Did you happen to measure the low frequency bandwidth? With the 300b in George's amp it was -3db at 11.7Hz and it's starting to roll off at 30Hz. The 300b has lower plate impedance so I figure the 6Y6 won't have the same roll off characteristic. My guess is it will be around -3db at 20Hz and start it's roll off much higher. Even with my cheapo OPT the bass on my full range mono setup wasn't bad, I only noticed how dull the recordings sounded which led me to look into the high frequency domain which showed I was losing the high frequency detail from the recording.

 

 

Did not bother this morning, but did one day last week. Looked good to me. Surprising to me is the excellent bass response for just one watt of power. Best some of my other amps with better numbers sonically. 

[captainbeefheart]

I just double checked and the high frequency roll off isn't from the output transformer so I do apologize for that assumption on my behalf.

 

The high frequency roll off is definitely from the circuit itself,  from the high output impedance of the 6SJ7 stage and the input capacitance of the 6Y6.

 

Of course layout will give some differences in stray capacitance, my breadboard had more than Henry's layout but it's no coincidence that we are getting very similar roll offs with completely different output transformers.

 

The sonic impact this has on fidelity is completely dependent upon how good your hearing is like Henry said. Most of the old timers may not even notice it while the younger guys with excellent ears will. I even noticed it right away but just thought it was my really crummy OPT was the cause.

 

I should have known better. I have made no feedback SET amps with two stages before and ran into the same problem with high impedance driver circuits. I eventually just stopped doing two stage circuits because with low powered SET amps I feel that Class A2 has a large advantage so I always have a buffer between the voltage amp stage and output triode. With Class A1, two stage designs I was using high gm pentodes like the 6EJ7 wired as a triode. It gave excellent gain, very low output impedance to drive the output triode input capacitance, and plenty of current to not run into Slew rate problems. With a 300b tube one may require 140v peak to peak to reach full power and so since slew rate is a large signal phenomenon the output triode grid capacitance needs to be charged up fast enough. Wimpy high mu triodes and sharp cutoff pentodes didn't have enough current to reach 140v peak to peak into 300pF at 20kHz. With the Sweetie and it being 1 watt there isn't that much swing and 1ma of current should suffice just fine for enough to drive 60v peak to peak into the 6Y6 grid capacitance.

[captainbeefheart]

19 minutes ago, henry4841 said:

Did not bother this morning, but did one day last week. Looked good to me. Surprising to me is the excellent bass response for just one watt of power. Best some of my other amps with better numbers sonically. 

 

Even my really cheapo OPT sounded fine in the bass region. Our speakers don't go very low and most of us use sub woofers anyway.

 

Often OPT that do very well in the bass region will be a trade off from the high copper giving poor high frequency extension. The higher the impedance of the load needs the worse this problems gets, hence why I like to choose tubes that have a low plate impedance where I can use a low load impedance. A 3k:8 transformer will  be much easier to get both good bass and treble response say compared to a 10k:8 transformer. It's still possible but the transformer manufacturer will need a more complex winding/layering design. This is how they get very expensive quickly if we want excellent bandwidth in both directions.

[captainbeefheart]

This is the Universal Stancor OPT I used. I think I paid like $15 each for several of them. I like to use them for push pull guitar amps.

 

Ya I know, it screams hifi 🤣

 

It's actually 18 watt push pull but Stancor says you can use it for a single tube in the datasheet which I don't recommend. They must have added a very small air gap for this purpose. I just set it up for 10k plate to plate load impedance and connected a 10M45s to the other plate tap and adjusted current to null DC component in the winding. Works just fine since it's Class A. The bass sounded decent actually. I tried it with the single plate connection and center tap as Stancor suggests but the bass was not so good. It works but the minimal effort to hook a current source up to the other side was only fair to assess the Sweetie as best as I could.

 

 

[henry4841]

I think the design, circuit, is fine. Any limitations is in the tubes themselves is my feelings. Exactly why I like simple two stage circuits. One gets to hear what the tube sounds like without, or at least little manipulation, of the sound. No simple simple 2 stage amplifier such as this is going to shine on the bench. They just for some reason sound gorgeous for many of us. I like what I see on the scope and on the Arta distortion analyzer software. In my opinion it is doing what it is supposed to do, sound sweet in the mids and on vocals especially. 

 

I have built 2 SE KT-88's since building this amplifier. I just recently had time to play with my new signal generator that has distortion numbers low enough to use Arta is the only reason I decided to put this amp back on the bench and do some testing. Passed with flying colors when I finally figured out how to get 1/4 watt on the output. The distortion number you see is 2nd harmonic, what one builds a SET amplifier for. There are ways to lower the 2nd harmonic as Captain has been discussing but I like hearing the charm of the 6Y6 tube as my Great Grandpa heard it in his day. I am moving on with another project now. I have a First Watt Aleph J I built in my test chassis that I built for testing first watt designs that I need to put in a proper chassis I bought many months ago. Built the PS section yesterday. Life is good. 

[captainbeefheart]

I'm glad Henry was helping with measurements because I was going to just blame the OPT for the treble roll off. Since Henry was getting the same roll off with the Hammond's I figured something else had to be going on which looking closer at the 6Y6 in triode mode's input capacitance and the source impedance of the driver stage the numbers fit perfectly. I even checked with a simulation since I dismantled the breadboard sweetie circuit.

 

Miller capacitance is rearing it's ugly head. There is enough gain to which one could try and change the 100k plate load resistor for the 6SJ7 to a lower value, maybe try 82k which should help lower the output impedance of that gain stage. This should bump the  corner frequency past 20kHz and only bump the open loop gain from 17db to 15.5db.

 

Since the limitation is in the circuit itself and not the output transformer than I change my advise to not spend big money on a high bandwidth output transformers.

[tube fanatic]

18 hours ago, captainbeefheart said:

 

I have excellent hearing for an old guy and I noticed it immediately. Which is why I added the feedback to increase the bandwidth.

 

 

 

 

Captain, at your age I would expect your hearing capabilities to greatly exceed mine and Henry’s.  Any time you want to trade let me know!  I’d gladly turn the clock back 30 years 😀!

 

Maynard