Ikezilla - 120 watt Ike?

a DIY, modified Mullard 5-20 monoblock design

Ikezilla - 120 watt Ike?

Postby Thermion » Mon Nov 14, 2005 10:05 am

I've been planning to build a pair of Ikes for a long time, but the project kept getting put on hold. Probably because I installed a pair of Shannon's prototype ST70 boards in my MKIII's and was so happy with the result I didn't need another pair of 60 watt PP UL KT88 amps. However, more power would get me excited.

So, I approached Shannon to see if he was interested in playing around with PPP UL KT88 designs. We pooled resources and threw together a prototype amp. It is based on the stock Rev A Ike circuit. In fact, we used an Ike board for the power supply, grounded cathode, and LTP. (12ax7 and 6SN7) I then wired up some chassis mount octals with flying wires to the octal footprint on the board. Also threw in a simple bias board to allow individual bias adjustment on the power tubes. The PT was from Mayland Toroid, http://www.toroid.com/standard_transfor ... s_tube.htm and the OT was a Hammond 1650T (1.9K 120 watt). The tubes were a well used quad of Sovtek KT88s.

This was a good learning experience. The first time we powered up, we had blue arcing in one of the tubes due to parastitic oscillations. We had to add plate and screen grid resistors to stabilize things.

Yesterday I took the prototype over to Shannon's and we tweaked the feedback and swept it to see how it looked. Open loop gain was just over 40 db. Here are 50 watt results with a B+ around 450 volts and a lean bias of 68 ma per tube.

20Hz 14.66V +21.34dB .746%THD+N 51.84W @ 4ohm 81.88dB S/N
25Hz 14.38V +21.19dB .464%THD+N 51.62W @ 4ohm 81.76dB S/N
30Hz 14.39V +21.19dB .406%THD+N 51.77W @ 4ohm 82.1dB S/N
50Hz 14.37V +21.19dB .355%THD+N 51.48W @ 4ohm 81.62dB S/N
100Hz 14.37V +21.20dB .292%THD+N 51.48W @ 4ohm 81.97dB S/N
200Hz 14.37V +21.20dB .292%THD+N 51.55W @ 4ohm 80.85dB S/N
500Hz 14.36V +21.18dB .304%THD+N 51.48W @ 4ohm 82.16dB S/N
1kHz 14.37V +21.20dB .297%THD+N 51.55W @ 4ohm 82.dB S/N
2kHz 14.37V +21.20dB .291%THD+N 51.55W @ 4ohm 81.43dB S/N
5kHz 14.32V +21.16dB .311%THD+N 51.19W @ 4ohm 81.89dB S/N
10kHz 14.18V +21.07dB .389%THD+N 50.2W @ 4ohm 81.18dB S/N
20kHz 13.72V +20.78dB .661%THD+N 46.92W @ 4ohm 81.08dB S/N
30kHz 13.03V +20.32dB .988%THD+N 42.32W @ 4ohm 80.99dB S/N

Our goal was to get 120 watts so we cranked up the B+ to 475 volts and biased the tubes to 81 ma. Here are the results.

20Hz 20.85V +20.77dB 18.316%THD+N 108.3W @ 4ohm 80.01dB S/N
25Hz 22.02V +21.25dB 1.882%THD+N 121.2W @ 4ohm 83.23dB S/N
30Hz 22.04V +21.26dB 1.142%THD+N 121.3W @ 4ohm 83.94dB S/N
50Hz 22.07V +21.28dB 1.312%THD+N 121.W @ 4ohm 84.2dB S/N
100Hz 22.07V +21.28dB 1.179%THD+N 121.1W @ 4ohm 83.98dB S/N
200Hz 22.08V +21.28dB 1.056%THD+N 121.2W @ 4ohm 83.94dB S/N
500Hz 22.05V +21.27dB 1.055%THD+N 121.1W @ 4ohm 84.06dB S/N
1kHz 22.07V +21.28dB 1.011%THD+N 121.3W @ 4ohm 83.83dB S/N
2kHz 22.07V +21.28dB 1.006%THD+N 121.3W @ 4ohm 83.89dB S/N
5kHz 21.99V +21.24dB 1.002%THD+N 120.5W @ 4ohm 83.81dB S/N
10kHz 21.79V +21.16dB .994%THD+N 118.5W @ 4ohm 83.91dB S/N
20kHz 21.06V +20.87dB 1.161%THD+N 110.5W @ 4ohm 83.48dB S/N
30kHz 19.94V +20.37dB 2.27%THD+N 99.1W @ 4ohm 83.02dB S/N

1% distortion at 1k at 121 watts!!! Yellow_Light_Colorz_PDT_10

Now that's something I can get excited about.

JT
Thermion
 
Posts: 181
Joined: Mon Jul 07, 2003 6:04 pm
Location: Fisher, IL

Postby erichayes » Mon Nov 14, 2005 11:23 pm

Hi All,

You guys are on the right track. I've never worked with Hammond iron in the outputs, but the 1650T seems to me to be a little lacking in mass for solid low end response, and the rolloff at 30kc seems excessive. Can you post some 1 watt figures with the amp running at 475 plate volts and 72mA/tube cathode current? If you're using the 8903, run 1kc first and go to relative dB and set to 0. That'll keep me from having to do a bunch of mental gymnastics when you run the other frequencies. 'preciate it.

A few suggestions:

Change the 6SN7 to a [12BH7 or 6GU7] with balanced sections.

Change the 12AX7 to a 5751.

If there isn't an AC balance control on the plates of the PI, add one.

If you're using half-wave rectification for the bias supply, go to full-wave.

If you're using gold plated tube sockets, replace them with regular tinned ones.

Make your measurements using an 8Ω load connected to the 4Ω tap on the OPT.

Bud and I are getting 111 watts @ 0.3% THD out of a pair of monoblocks that started out as essentially the same platform y'all are using. The iron's proprietary, of course, as is the circuitry beyond the level I've outlined above. The point is: it can be done, and when you get the blood scent, God help your spouses.
Eric in the Jefferson State
erichayes
KT88
 
Posts: 987
Joined: Fri Jan 23, 2004 9:01 pm
Location: McKinleyville CA

120W Tweaking

Postby Shannon Parks » Tue Nov 15, 2005 7:40 am

Hi Eric,

erichayes wrote:I've never worked with Hammond iron in the outputs, but the 1650T seems to me to be a little lacking in mass for solid low end response, and the rolloff at 30kc seems excessive.


Since these are initial closed loop numbers with aggressive 'slugging capacitance' in there, this data is a little skewed per the >20kHz evaluation. In my limited experience, for stability on whatever type of loudspeakers used I like to see flat from 20Hz to 10kHz with about a 1/2dB dip at 20kHz, so that is what we shot for. I imagine Thermion is charting all the different data using the software he is comfortable with and he'll post a lot more, particularly the open loop data at 1W. As for the low end, the 1650T isn't even as big as the HF89's iron, so we definitely are on a primary H budget. Yellow_Light_Colorz_PDT_05 Still need to do square wave testing at 10kHz and tune out the ripple with a step circuit before the LTP. Probably re-adjust the global feedback combo at that time.


erichayes wrote:Can you post some 1 watt figures with the amp running at 475 plate volts and 72mA/tube cathode current? If you're using the 8903, run 1kc first and go to relative dB and set to 0. That'll keep me from having to do a bunch of mental gymnastics when you run the other frequencies. 'preciate it.


I'll try and revamp the diytube AA software to version 2.0, incorporating the log excel charts that I used for the ST70.

erichayes wrote:A few suggestions:

Change the 6SN7 to a [12BH7 or 6GU7] with balanced sections....If there isn't an AC balance control on the plates of the PI, add one.


Cool - I love the 12BH7/6GU7, too - very linear! I think we are stuck with octal for the short term, though. What about 6BX7/6BL7? Would these be a step back from the 6SN7 in this application? But first, we've already decided to change the stock LTP to using a CCS (the ubiquitous LM334Z). Hopefully that will give us an improvement on the balance, though adding a balance pot wouldn't be difficult.

erichayes wrote:If you're using half-wave rectification for the bias supply, go to full-wave.


It is full wave (thanks to the dual bias taps from the barely warm Toroid of Maryland). I'm not sure if it is filtered enough. What is a good target of ripple (in uV or mV) in a bias supply?

erichayes wrote:If you're using gold plated tube sockets, replace them with regular tinned ones.


Cheap PCB ones for the 12AX7 and 6SN7 and NOS phenolic for the KT88s.

erichayes wrote:Make your measurements using an 8Ω load connected to the 4Ω tap on the OPT.


What is the purpose of measuring like this?

erichayes wrote:Bud and I are getting 111 watts @ 0.3% THD out of a pair of monoblocks that started out as essentially the same platform y'all are using. The iron's proprietary, of course, as is the circuitry beyond the level I've outlined above. The point is: it can be done, and when you get the blood scent, God help your spouses.


O man, preach it brother! They'll be ticked when they see the bill. Yellow_Light_Colorz_PDT_13 Yellow_Light_Colorz_PDT_04

Thanks for the tips and finally whetting our appetitites enough get started on this!

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby dhuebert » Tue Nov 15, 2005 8:44 am

I would appreciate graphical representation, rather than columns of numbers. Helps me picture things.

Don
User avatar
dhuebert
KT88
 
Posts: 820
Joined: Thu May 01, 2003 9:26 am
Location: Winnipeg Manitoba Canada

Postby Thermion » Tue Nov 15, 2005 10:41 am

Eric,

Thanks for the comments, Shannon did a pretty good job at answering line by line, but here is some of my own thinking on the subject.

The Hammond outputs are a little lacking in iron, they only spec them down to 30 Hz. Not much musical information below that anyway, plus very few speaker systems with response down there. I’m using a pair of 1650Ns in my Dyna MKIIIs and been real happy with the sound. I bought the 1650T because it was cheap and readily available. Send me a PM if you’re willing to sell a pair of yours for the production models.

I want to stick with octal tubes for the amp. I have a bunch of closely matched 6SN7 pairs, but very few 12BH7 or 6GU7. We may try some 6BX7s.

I plan to try a 5751 next, but probably for different reasons than you think. In the production pair I am going to use Rev B Ike boards with 6SL7 types for the GC stage. The 5751 is closer electrically to the 6SL7 than a 12AX7.

I’m not planning to spend time charting this preliminary data. It is the first look Shannon and I have on the performance. We are going to play around a little and see what we can attain. Then I’ll produce some pretty graphics for everyone to look at. Maybe we can get Shannon to post a couple of photos too.

JT
Thermion
 
Posts: 181
Joined: Mon Jul 07, 2003 6:04 pm
Location: Fisher, IL

Postby erichayes » Wed Nov 16, 2005 5:09 pm

Hi All,

The first versions of my 50/50 watt and 100 watt amps were all octal (common 6SL7, [6SN7, 2 x KT88/chan] for the 50/50, 6AQ7, 6SN7, 4 x KT88 for the 100), and they sounded pretty damned good, with distortion figures in the mid 0.0X% range at 1 watt, and frequency response flat from 10~100,000 cps ±1dB. When Bud got here and we decided to push them to the extreme, the first thing he wanted to do was get rid of the SL and SN and essentially transplant the 1773's front end into them. Reluctantly, I made the changes to one of the 50/50s, leaving the other one original for comparison purposes.

Well, there wasn't any comparison--we didn't bother firing the stock unit up. Although the distortion and frequency response measurements improved modestly, the transient response and overall ability of the amp to deal with complex orchestral works was astounding. When I expressed my mixed feelings over the loss of the cool looking octals, his terse response was "Whatterya gonna do? Look at it or listen to it?" End of that subject. . .

When Terman, Langford-Smith et al came up with their output loading formulae, they had overall efficiency vs minimum distortion in mind as their goal. Fidelity literally didn't enter into the equation. We (and others) have found that, if the amp is loaded with around twice the "normal" impedance (i.e. 8Ω working into the 4Ω tap of the OT), the output tubes can operate in a much more relaxed environment, causing sound quality to improve and distortion to decrease, at the cost of a few dB in output. But, you say, those are different terms for the same thing. No, they're not; sound quality is what lets you keep listening to an amplifier for hours without ear fatigue, while distortion is any deviation from the original sound introduced by a reproducing component. The two go hand in hand, but we've all heard amps that had the same specs and sound totally different from one another.

The bias supply I've used in all the amps is a FWB working into a pi filter with 3.3K between a couple of 100 µF caps and to a voltage divider to get the voltage close to what's needed. Typical ripple is around 3.5 mV RMS.

Finally, it's probably the company I've kept earlier this year, but I've learned a couple of things about myself: first, I can hear a hell of a lot more than I've ever given myself credit for. I just had to learn to listen. Second, I don't have to hide behind apology, rationalization or compromise where sound is concerned--I can make it better. This is not Tony Robbins hype/BS, but fact. Our amps are running 10~100,000cps ±0.2 dB @ 10 watts, have <.03% THD and IM @ 10 watts, can dump >10 watts into 2.2 µF with a barely perceptable increase in ringing on a 10 kc square wave, and are not ashamed to play LOUD

When I was in retail back in the '70s I used to joke that audio is a lot like astronomy: you think you're happy and satisfied until your friend brings over a bigger telescope. . .
Eric in the Jefferson State
erichayes
KT88
 
Posts: 987
Joined: Fri Jan 23, 2004 9:01 pm
Location: McKinleyville CA

more tips

Postby Shannon Parks » Thu Nov 17, 2005 7:26 am

erichayes wrote:Second, I don't have to hide behind apology, rationalization or compromise where sound is concerned--I can make it better. This is not Tony Robbins hype/BS, but fact. Our amps are running 10~100,000cps ±0.2 dB @ 10 watts, have <.03% THD and IM @ 10 watts, can dump >10 watts into 2.2 µF with a barely perceptable increase in ringing on a 10 kc square wave, and are not ashamed to play LOUD


This is pretty phenominal. I'm just happy to have a stable amp with a 2.2uF capacitive load. Looking forward to further info about your amp! Thanks for setting the benchmark high so we won't settle.

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby TerrySmith » Sat Nov 19, 2005 9:27 am

Any chance of an Ikezilla board?

This reminds me of when I swapped a Chevy 327 with tuned port injection into a Fiero! Lots of power- YES! Overkill- not really.
T. Smith
User avatar
TerrySmith
KT88
 
Posts: 973
Joined: Sat Jan 31, 2004 12:51 pm
Location: Maryville TN

Postby WA4SWJ » Sat Nov 19, 2005 9:56 am

As someone said:

"You can never have too much horsepower."

How 'bout some pictures of the ikezilla?

Regards,
Ed Long
User avatar
WA4SWJ
KT88
 
Posts: 650
Joined: Mon Dec 27, 2004 8:39 pm
Location: Belleview, FL

My best copy of Audio Precision...

Postby Shannon Parks » Tue Nov 22, 2005 7:39 pm

dhuebert wrote:I would appreciate graphical representation, rather than columns of numbers. Helps me picture things.

Don


Here 'tis, Don - my best Audio Precision knock-off:

Image

Nevermind that these numbers from Thermion's original post are history. Yellow_Light_Colorz_PDT_10 Did some work the other night and we have it well on its way to serious tube smackdown, eg 150W at 1%THD+N 1kHz. Will try to post more numbers & actual pictures (it's a mess right now with test leads everywhere) in a couple weeks. Holidays are leeching tube time!! Yellow_Light_Colorz_PDT_45

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby dhuebert » Wed Nov 23, 2005 8:25 am

I hate to be a pest, but...
Nice job on the graph, but can I have another? How about distortion @ 1KHz from 0.1Watt to say 5% THD (130 Watts?)? I just bought all the parts to stuff my ikes from Roy and a quartet of whatevers sounds like just the ticket.

(i.e. 8Ω working into the 4Ω tap of the OT),


I increased the power of my bass amp (quartet of 6L6s) from 55 watts to 68 watts by using an 8 ohm load on the 16 ohm tap (from 8 ohm load on 8 ohm tap). This reduces the load on the tubes by half since the plate resistance is half due to parallel. The reason I went to quartet was to double damping factor, better to control speaker excursions for those snappy passages, rather than for high power. Question is: what happened to damping factor when I halved the load on the tubes?

Don
User avatar
dhuebert
KT88
 
Posts: 820
Joined: Thu May 01, 2003 9:26 am
Location: Winnipeg Manitoba Canada

Postby Thermion » Wed Nov 23, 2005 11:00 am

dhuebert wrote:I hate to be a pest, but...
Nice job on the graph, but can I have another? How about distortion @ 1KHz from 0.1Watt to say 5% THD (130 Watts?)? I just bought all the parts to stuff my ikes from Roy and a quartet of whatevers sounds like just the ticket.

Don


Don,
I’m at a loss to figure out why you want this information in this form. What purpose do you have in mind? Why only the 1KHz information? As you know the THD figures are highly dependent on the circuit, the OT and the tubes used.

Shannon and I would be happy to try and accommodate your request, but some additional information would help us keep from wasting time. We are having trouble finding the precious hours to work on this, and we are still tuning the circuit and trying out other circuit modifications.
JT
Thermion
 
Posts: 181
Joined: Mon Jul 07, 2003 6:04 pm
Location: Fisher, IL

Postby dhuebert » Thu Nov 24, 2005 8:27 am

Gee, I'm not sure what to say.. Don't we always see this info for amplifiers? Distortion VS frequency and distortion VS power @1KHZ are pretty standard for describing amplifier performance. Having said that, I don't want you wasting alot of time on this, I'm just curious. In my days as a number cruncher it would take as long as punching in the numbers to create a graph in Lotus 123. I looked at the numbers you posted in the forum and thought why not spend the time making a graph in excel and cut and paste to the forum. Maybe there is something here I'm not aware of that makes it quicker to punch a bunch of numbers into the forum rather than draw a graph. Anyways, I am still very interested in the project, keep us posted!

Don
User avatar
dhuebert
KT88
 
Posts: 820
Joined: Thu May 01, 2003 9:26 am
Location: Winnipeg Manitoba Canada

AP Clone

Postby Shannon Parks » Sun Nov 27, 2005 12:27 pm

dhuebert wrote:Gee, I'm not sure what to say.. Don't we always see this info for amplifiers? Distortion VS frequency and distortion VS power @1KHZ are pretty standard for describing amplifier performance.


Hi Don,

Don't worry - I'll revamp my test software this winter and get more & better data than ever. I'm glad to have your interest!

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

My Bad!!

Postby Thermion » Sun Nov 27, 2005 2:36 pm

dhuebert wrote:.. Don't we always see this info for amplifiers? Distortion VS frequency and distortion VS power @1KHZ are pretty standard for describing amplifier performance.
Don


Don,

My bad on this one. It occured to me what you were asking for after I posted. I had a senior moment and didn't see the forest for the trees. (Pretty sad for someone whose day job is analyzing experimental data and putting it into graphical form.) Yes, the 1K distortion versus power plot is pretty much standard. At this stage, we only have a couple of points to plot, so it wouldn't be very interesting. As we get the circuit tuned, (and swap in the 5751) we'll put post the 1K plot and lots of other information.

But since you brought it up, I have never found the 1k plots to be useful to describe amplifier performance. You can find the "knee" and determine when transistors saturate. With tubes its not as clear. For our purposes, it tells you if the amp is performing correctly, but gives no information about how it will sound, or how it will handle a reactive load. Anyway, I find the distortion versus frequency and gain versus frequency for various power outputs to be far more telling of an amps overall performance.

JT
Thermion
 
Posts: 181
Joined: Mon Jul 07, 2003 6:04 pm
Location: Fisher, IL

Next

Return to eiclone

Who is online

Users browsing this forum: No registered users and 8 guests