AudioCircle
Audio/Video Gear and Systems => The Lab => Topic started by: Paul Hynes on 28 Mar 2009, 04:11 pm
-
Hi Folks,
A long time ago, when the LAS (Line Array Speakers) circle was live, I offered to describe two simple power output stages, capable of driving low impedance loads, such as parallel wired line arrays or ribbon drivers. There was little interest shown at the time, and the LAS circle was archived, before I was able to find the time to prepare the information.
There has recently been a resurgence of interest in this type of design and several AC members have asked me to describe these output stages for them to try. These output stages are adaptable to any loudspeaker load, not just low impedance loads, and I have made versions up to 100 watts into 8 ohms by adding voltage gain stages constructed with valves and resizing the power supply voltage/current capability. I will present designs for voltage gain stages if any interest is shown on the forum.
Firstly, for those new to this subject, I will recap on my usage of the output stages in my system. I use an open baffle line source array loudspeaker design in my system. After conducting listening trials, I concluded that it was important to wire all the drive units of an array in parallel, because if you wire them in series/parallel, the amplifier cannot control the drive units properly. Think about it, if you wire loudspeaker drive units in series, you are introducing a highly reactive electrical spring, between the amplifier and the drive units.
I use 8 Visaton B200 drive units in each line array. Wired in parallel, this presents a nominal 0.75 ohm load to the amplifier. Most normal amplifiers do not like this so I had to design an amplifier specifically for the job. In fact I designed two output stages, one was single ended (SE1) and one was based on the “circlotron” (CIRC1) output stage. My line array is very efficient with a sensitivity of 105 dB for 1 watt at 1 metre, so I do not need a lot of power, or any additional voltage gain in the power amp, to generate realistic sound pressure levels. For this reason I use the single ended version in my system and I also happen to prefer this marginally for it’s warmer tonality. Both designs can be used in class A and the CIRC1 can also be used biased in class AB allowing much higher power outputs than the SE1.
The circuit of the single ended version (SE1), is at :-
http://i700.photobucket.com/albums/ww7/paulhynes/SE1.jpg
The SE1 follows the KISS principal by only having a resistor, a mosfet, the power supply and some wire in the signal current path. It is DC coupled and if you use matched mosfets for T1 and T2, mounted close together in the centre of the heatsink, their on resistance will track very well, minimising offset drift. These type of circuits have limited power supply rejection so a good regulated supply is mandatory with this circuit.
The circuit of the “Circlotron” version (CIRC1), is at :-
http://i700.photobucket.com/albums/ww7/paulhynes/CIRC1.jpg
The CIRC1 circuit will only accept a balanced input signal for proper operation. It is important to note that the CIRC1 power supplies must be floating with respect to earth as they derive their reference from the source terminal of the opposing mosfet device. Both mosfets should be mounted close together on the same heatsink for good thermal tracking to minimise output offset voltage.
The depletion mosfets, used in both designs, allow very simple bias arrangements to be used, but as shown they will run at IDSS, which, in my system application is around 2.5 amps at their working temperature. Decent heatsinks are mandatory.
Both designs benefit from quality power supplies. I am running both designs on 12 volt supply rails and the SE1 will develop approximately 2.5 watts into 0.75 ohms. The power delivery is limited by the Class A current available. I could actually reduce the supply rails for the SE1 to around 4 volts, and still be able to deliver the 2.5 watts. Doubling the number of mosfets and running on 7 volt supply rails would allow an output of 20 watts into 0.75 ohms. The CIRC1 will develop about 60 watts with 12 volt supplies as long as the power supply can provide the required 13 amps current, as technically it is biased in Class AB but with a high Class A standing current. It is possible to adjust the bias on the CIRC1 to a lower level allowing a much more efficient output stage. In theory it should be feasible to design a kilowatt output stage using multiple parallel mosfets but I am not going there at the moment, as I don’t need to.
There is another big advantage with high efficiency parallel-wired line source arrays that is often overlooked. Because you need so little voltage swing from your amplification, you can reduce the circuit clutter in the signal path to very little. For example the output from the Altmann Dac or my phono stage that I am currently using is more than sufficient to achieve the sound pressure levels I require with no further voltage amplification at all. It just feeds into my Lightspeed volume control and straight into the SE1 current amplifier. Because of this simplification, both these designs offer a high level of transparency to the music. If you have a dac with balanced outputs you can do the same with the CIRC1, however you would need a single ended to balanced converter of some description if your dac is single ended out.
Some of you who want to try these designs will need to adjust various parameters to suit your system and I will be happy to help with advice. Just ask.
Regards
Paul
-
Very cool. I'm always interested in new topologies or ways of accomplishing things I have not seen before.
The photo links were to the thumbnails. Full size links:
http://i700.photobucket.com/albums/ww7/paulhynes/SE1.jpg
http://i700.photobucket.com/albums/ww7/paulhynes/CIRC1.jpg
-
Thanks Mike,
I haven't used photobucket until recently and I should look at the instructions more closely. I will paste the correct address into my post.
Regards
Paul
-
Paul can you tell us yr story how you experience your B200 array?
In the end this is what you have designed your current amps for.
I would enjoy yr listening revies of your speakers
incl source and rest of setup?
-
Hi Kyrill,
I didn’t design the “Circlotron” output stage specifically for the B200 array. I have been using a version of this output stage with positively biased lateral mosfets for over 10 years. Specifically a version designed to deliver 100 watt into 8 ohms, with a tube differential pair driver stage.
When IXYS launched their high power depletion mosfets, they actually answered my prayers, as I had wished such a device existed many times in my design career. I have been using Siemens low power depletion mosfets, for various functions, since they were launched, in the nineteen eighties. In fact I built a low power “Circlotron” with many of these in parallel first but decided to drop this design when IXYS produced their depletion power mosfets. They just made life so much easier as they are very rugged, easy to apply, and the sound system sounds very good with them in the output stage.
When I designed the B200 array I had a hunch that parallel wiring would bring useful improvements to the performance so I had to re-design the power amp output stages to cope with the low impedance load. It was relatively easy as all that was required was an increased power supply current capability and reduced supply rail voltage to keep dissipation under control. This was possible because low impedance loads do not need a lot of voltage swing into the load to develop quite a lot of power.
I tend to prefer single ended designs to balanced designs for their tonal character so it wasn’t long before the IXYS devices were in my system as the SE1 topology set up to deliver 10 watts into 8 ohms. I liked it so much it has stayed there even though it only delivers a maximum of 2.5 watts into 0.75 ohms. I just love high efficiency loudspeakers. I have toyed with uprating the SE1 to 20 watts onto 0.75 ohms but in practice I do not actually need the additional power in my current listening room. I also like the Circlotron output stage although it would be described as more towards a neutral tonal balance although the difference is not dramatic. Probably because it is made from two SE stages wired in the "Circlotron" topology.
It has to be said that, with such simple circuits, the power supplies contribute considerably towards the overall performance quality, but get it right and the designs sing with great beauty on all types of music.
I have run out of time as there is some family stuff to do so I will continue with some listening notes on the line array/current amps when/if I get a moment tomorrow.
Regards
Paul
-
Hi Paul,
now we only need 16xB200...
what did you mean by driving this amp with the altmann byob?
LS output to in of this amp?
love the simplicity...I also made the biggest improvements by simplifying things in my previously built stuff...
kind regards,
Tony
-
Paul, can you give us transformer VA to size the power supply? Thanks for the circuit.
-
Circlotrons are so sexy! Full range single driver speakers prefer current amps.
Thanks Paul!
-
Hi Tony,
If you like the abilities of the B200 you will not be disappointed and you certainly will not go back to a single pair of B200 as you will loose too much.
The B200 is not a cheap drive unit (it retails at nearly a hundred pounds in the UK) and the cost of 16 of them can be a little daunting in the current financial climate. For those with budget limits who have not tried the B200, I suggest getting one pair and fitting them to a baffle. Let them run in for a month or so to loosen up and if you like what you hear and want to go for the line start by preparing the line baffle and a sturdy frame to hold it as rigid as possible. Then as funds allow you can add drive units until completed. Each time you add a pair of drive units you will get an upgrade towards the bigger picture.
Tony, the BYOB is a bridge amplifier with plus phase and minus phase outputs so you already have a balanced signal to feed CIRC1. You will need DC blocking capacitors as both the BYOB output phases are at half supply voltage under no signal conditions. Fitting a high quality dummy load resistor of say 8 ohms across the BYOB output terminals will keep the BYOB happy and the balanced signal just goes via the coupling capacitors to the CIRC1 inputs.
I like simple too. I have trod the complicated path many moons ago and have always return to the simple path to get to the music.
Hi Mike B.
The transformer VA size required will depend on the power requirements and set up of the amplifier with a given loudspeaker load. Let me know what you are going to drive and what sound pressure level you are looking for and I will be able to advise you.
Hi richidoo,
The CIRC1 does work well with the B200. It is pretty rugged too. God bless IXYS, they do make good devices, not cheap, but good.
Hi again Kyrill,
You asked about my listening experiences with the B200 array.
First, it may help to understand what I look for in a sound system. I have been involved with live music most of my life and I have played guitar in pop/rock groups, installed and worked in recording studios, acted as roadie for acoustic folk bands, designed valve amplification for musicians, designed sonically invisible PA systems specifically for acoustic folk music and regularly attended classical music concerts. I always pay particular attention to any acoustic instrument being played live even if it is simply a couple of guys busking in the street. This is because my reference for music reproduction is always live music.
Because of my awareness of the sound of live music, my sound system must create the illusion that real people are playing real instruments in a real acoustic space to give me any lasting satisfaction. I have been backwards and forwards on this path over the years just like the rest of you. The B200 array with the “Lightspeed” volume control (thanks George) and the SE1 or the CIRC1 is the closest I have been so far. This combination is effortless, dynamic, transparent and very musically involving at any level I care to play music from a whisper to a crescendo. I am a happy man.
Other equipment I am using at present includes :-
For analogue playback the Nottingham Analogue Spacedeck, unipivot arm and MM cartridge with my own design of valve phono stage. I also have a modified Systemdeck IIX with an Alphason Zenon arm and a Denon 103 MC.
For digital playback, whilst I work on building a new dedicated silent music server for use with my SB3 (it will also be set up as a ram player using Cplay), I am temporarily using the DVD player feeding the Altmann attraction DAC. The Jisco function is very helpful here in getting a good musical performance from what is basically a budget DVD player (although it does have the same DAC chip as the SB3). I had hoped to have all this sorted out by now but I have just been too busy at work to find the time recently. My daughter has expressed an interest in this project so we are going to build it together in whatever spare time we can find. Needless to say I will be applying my best power supply designs in the project.
Regards
Paul
-
This is somewhat off subject and I apologize in advance. I would like to reiterate that Paul over the past year has been very gracious with his time and designs some of the finest shunt regulated power supplies in the business. Now he is coming up with cool amplifier designs and has promoted the benefits of the Lightspeed volume control as well. If Paul has the time and energy to keep up with the likes of the diy'ers and non diy'ers in Audiocircle, it would be great if he has his own circle. So we should all keep that in mind and hopefully with an overwhelming vote, he can have his own circle.
His shunt regulator in my Poseidons Voice Katalyst power supply is absolutely fantastic, rock solid at 5V DC. And, currently, my SB3 is being upgraded by Wayne to implement Paul's 3 terminal regulators so there is a separate 3.3V linear supply to the DAC chip itself.
All in all I see him as an innovator who needs to be right here with us in our diy journeys.
Anand.
-
Hi Anand,
Thank you for the kind words. I'm getting embarrassed. I'm only trying to help people enjoy their music. There are plenty of others on the forum doing this as well and I have learnt from them. It is a two way thing.
Regards
Paul
PS glad you like the shunt reg. I was impressed with your build quality.
-
Thanks for the offer Paul. I will be driving a line consisting of 10 8 ohm (RE 6 ohm) mid woofers and 7 8 ohm (measure around 6) planar tweeters for each channel. They are currently series/parallel and about 95db efficiency. Average listening SPL high 80's rarely peaks over 100 db
-
Hi Mike,
“Thanks for the offer Paul. I will be driving a line consisting of 10 8 ohm (RE 6 ohm) mid woofers and 7 8 ohm (measure around 6) planar tweeters for each channel. They are currently series/parallel and about 95db efficiency. Average listening SPL high 80's rarely peaks over 100 db”
8 watts into 6 ohms should give you a little headroom for around 104 dB before clipping. I will assume you are staying with series/parallel wiring for the moment. If you are planning to go with parallel wiring I will re-calculate for you.
You can use either the SE1 or the CIRC1 as shown in the circuit diagrams I posted. Remember that the CIRC1 will need a balanced signal to work. Each power supply should be set to 12 volts and should be regulated with a regulator that has a fast transient response and very low impedance throughout the audio bandwidth. This will ensure that you have very low ripple and noise breaking through into the signal from the power supply. A typical DC voltage to feed the regulator would be around 4 volts higher to stay away from the dropout area. A 12 vac transformer winding would probably be fine if you use low voltage drop rectifiers and don’t have too much mains voltage variation. If you suffer from mains voltage droop go for a 15 vac winding and suffer the additional heat-sinking that will be required. Remember that this is per supply rail per amplifier. In other words you will need four regulated supplies for a stereo amplifier and they will all need to be floating with respect to earth as they take their reference from the current amplifier circuit which in turn takes it’s reference from the preceding equipment. Note that the CIRC1 power supplies are fully floating, as their reference is the source terminal of the opposing phase mosfet. In the SE1 circuit keep the wire between the two power supplies as short as possible and connect all the wires that reference to this junction in the dead centre. This will prevent wire impedance from creating reference errors.
At the working temperature the IXYS devices will have an IDSS of approximately 2.5 amps. This will ensure you stay in class A through the full working range of the current buffer and give some current headroom for reactive loads. Each IXYS device will dissipate 30 watts under static conditions. You should mount both IXYS devices centrally on the heatsink for that channel so they thermally track and they will need insulating from the heatsink. Make sure the thermal resistance of the insulator is low. If you use a 0.5 degree C per watt heatsink for each output stage the heatsink temperature will rise to approximately 30 degrees above ambient. The ideal place to mount the regulators is next to the output stage and this will also affect the size requirement of the heatsink. If you do this I would suggest that you go for a heatsink rated at 0.3 degrees C per watt for a temperature rise of around 32 degrees above ambient.
The minimum transformer VA rating per winding should be 15 vac at 7 amps AC giving 100VA. A 12 vac transformer would work out at 84 VA per winding. You can, of course increase this rating if you wish. If you build the CIRC1 and want to try parallel wiring you will need a bigger transformer but everything else will be the same.
If you have any further questions ask away.
Regards
Paul
-
Thank you for the information Paul. I checked the on line catalogs at Mouser and Digikey and they don't offer the IXTH 20N50D. Since they are the big IXYS distributors in the US, I am hoping a bunch want to build this amp and hound them about it.
-
Hi Mike,
I have noticed that this device is not a normal stock item with UK catalogue mail order companies. I had hoped this situation would be different in the States. I had to order mine from GD Rectifiers who are an IXYS agent in the UK. I was told that the minimum order was 30 pieces so it may be that a group buy is the best way forward. The devices are current production so if you have any trouble getting hold of them, get back to me, and I will organise a group buy.
Regards
Paul
-
Here is the reply from Digikey, which is a big domestic supplier
Thanks for the inquiry; part number IXTH20N50D is a non-stocking item for
us. The minimum qty is 60 pieces with a lead time of 8 weeks from time of
order. The unit price is $6.77283 each.
Also see our award winning website www.digikey.com for up to date pricing
and availability.
Looks like we are stalled until enough people what to try the circuit.
-
Hi Mike,
I will check with the UK distributors on price and availability for the IXTH20N50D.
How many do you need?
Regards
Paul
-
I will build a stereo amp, so four. How many did you need to match four?
-
Hi Mike,
I didn't match mine, I just used devices from the same batch. The offset was mv.
I will see what I can do for you later this afternoon when I have some urgent work finished.
If you do not mind adding a bias supply you can use Exicon Lateral Mosfets from Profusion PLC in the CIRC1 topology. They will ship small quantities from the UK. These give good sound quality also but with these devices I would recommend using the TO3 can versions as the plastic package versions can be a little sensitive to circuit interaction. The TO3 can provides a bootstrapped screen for the gate and drain sections on the die reducing their susceptibility to stray coupling. This does not happen with the IXYS plastic package devices I have tried. If you decide to go this way I can post a suitable circuit showing the bias supply.
Regards
Paul
-
This is somewhat off subject and I apologize in advance. I would like to reiterate that Paul over the past year has been very gracious with his time and designs some of the finest shunt regulated power supplies in the business. Now he is coming up with cool amplifier designs and has promoted the benefits of the Lightspeed volume control as well. If Paul has the time and energy to keep up with the likes of the diy'ers and non diy'ers in Audiocircle, it would be great if he has his own circle. So we should all keep that in mind and hopefully with an overwhelming vote, he can have his own circle.
His shunt regulator in my Poseidons Voice Katalyst power supply is absolutely fantastic, rock solid at 5V DC. And, currently, my SB3 is being upgraded by Wayne to implement Paul's 3 terminal regulators so there is a separate 3.3V linear supply to the DAC chip itself.
All in all I see him as an innovator who needs to be right here with us in our diy journeys.
Anand.
+1 :D
Yes, it would be great for Paul to have his own Circle to share his wisdom and experience with us!
:thumb:
Best Regards
Dennis
-
+1
-
Hi Folks,
For those interested in building the current amplifiers SE1 or CIRC1.
I have checked with one of the UK distributors of IXYS Semiconductors and they will supply me with a smaller quantity than Mike was quoted from USA distributors. Unfortunately the price is not comparable to ordering 60, but who wants to order 60 when you only need 4.
After covering my costs and the time element of order processing I can offer the IXTH 20N50D at £12 each plus insured carriage and packing at cost. I will be ordering some for myself, during the next week, so if anyone is interested, let me know and I will include them in the order. I have been told I can have a small order shipped within 14 days at present as they can access some stock fairly quickly. I will work out the carriage/packing cost tomorrow and post the details.
Regards
Paul
-
Hi Paul,
I have picked up on this from your post on DIYA. Thank you for introducing us to an innovative & simplistic & more importantly good sounding circuit. It's these moments that keep me coming back to to the DIY forums - you never know when or where these unique ideas might spring up!
Not wanting to drag the thread off in another direction but you mentioned that lateral mosfets may be used. I have some of these lateral mosfets (12A 200V 125W RMS) http://www.classd.ltd.uk/product.php?productid=16174&cat=0&bestseller=Y which come at a very keen price of £3.30 & I thought of using these. There are also double die versions (24A 200V 250W RMS) available for £7.48.
Could you post a suitable bias circuit, please. Any other considerations need to be taken into account in using these?
I'm also wondering if the added cost & complexity of the extra bias circuitry needed for these parts is worth the few bob in savings?
I'll no doubt have some Qs about the PS when I study some more - if that's OK?
-
I would be interested in a group buy - only need 15 people at 4 devices each for the minimum order qty of 60!
I don't have high efficiency speakers. My two current sets are both mini-monitors - Jordan JX92S & Rogers LS3/5A - not efficient by any means so I need a voltage gain stage. I'm building one based on a beam deflection tube circuit by John Swenson which gives a balanced out signal. This, I imagine, would be perfect for use with the Circ1.
-
Hi JK,
The CIRC1 circuit with Lateral Mosfets works very well. I would use the double die version but in a metal can if you can get these. This will give plenty of reserve to drive difficult loads. The bias circuit (positive for the lateral mosfets) is quite simple and can be driven from the HT in the voltage gain stage. I will prepare some documentation for you when I have finished posting the "Lightspeed" remote control documentation on my website. This is nearly ready now as it just needs tidying up and including in the updated website which I should have posted over the next day or two.
With low efficiency 8 ohm speakers you will need higher supply rails than the amplifiers I use to drive the line arrays so you would have to use a bias circuit on the IXYS devices (in this case a negative bias just like a valve) to drop the quiescent current to limit the dissipation to sane levels.
How much voltage gain will the Swenson circuit give you?
Regards
Paul
-
Hi Paul,
I'm looking forward to your VCCS - I'm on the prepaid list for these.
John Swenson has had this circuit up to 90V PP without any increase in distortion. Schematic here: (http://johnswenson1.home.comcast.net/~johnswenson1/stereo/BDTpreamp.gif)
Without giving away your commercial secrets, any pointers to a good regulated high current PS schematic?
-
BTW, this BDT preamp is like the lightspeed in that the signal doesn't go through a potentiometer. The volume is controlled by adjusting the voltage on the screen & as such is also open to remote control
Edit: The Jordans are 8 Ohm, I think but the Rogers LS3/5As are nominal 15 ohm speakers - are there other considerations for a 15 ohmer like this?
-
Hi JK,
Sorry I can't release any power supply schematics as they are either in use commercially and/or the subject of non-disclosure agreements.
I can recommend a regulator module for the HT supply for you, but I had better not do that on this thread, as I would break the commerce rules. It will be tomorrow before I can get back to you as I will be busy for the rest of the afternoon preparing for a director's meeting this evening. I will also deal with your post about loudspeaker impedance tomorrow.
Regards
Paul
-
No problem, Paul
-
Hi JK,
The Rogers LS3/5a is very insensitive at 83 dB for 1 w at 1 mtr and with limited power handling (only 25 w continuous with the earlier models). I used these speakers extensively in the 1980's so I am aware of their quirks. I would aim for around 50 watts into 15 ohms nominal for these speakers and forget high level listening. This will get you to around 100 dB before clipping.
The CIRC1 configuration on 45 volt supply rails using lateral mosfets biased just high enough to remove crossover artifacts would work well with the LS3/5a and would allow 100 w delivery into the 8 ohms Jordans. The sensitivity of the Jordans is higher at 88 dB for 1w at 1 mtr and the power handling is 50 w continuous. 100 w should take you to around 106 dB before clipping.
If you are worried about exceeding the power handling of the speakers you would be more likely to damage them by repetitive clipping of a smaller amplifier (this can generate a surprising amount of energy that can overheat the driver voice coils) than occasional clipping of a larger amplifier.
The amplifier power supply should be rated to delivery at least 5 amps for 100w into 8 ohms and more if you intend to use it with lower impedance loads in the future. Because this application will use 45 volt approx supply rails It would not be sensible to use it on very low impedance loads like my line arrays. 4 ohms would be ok.
I posted about a regulator option on the "Paul Hynes Power Supplies" thread in the "Industry Talk and Events" circle.
Regards
Paul
-
Paul,
I saw the recent posts by you and Sondale on your SE1 and was very intrigued. I have very inefficient planar speakers, but Sondale's speakers seem to be about as in-efficient and the SE1 works well for him... so now I'm very interested.
In your first post in this thread, you said:
"These output stages are adaptable to any loudspeaker load, not just low impedance loads, and I have made versions up to 100 watts into 8 ohms by adding voltage gain stages constructed with valves and resizing the power supply voltage/current capability. I will present designs for voltage gain stages if any interest is shown on the forum."
Could you please?
THANKS!
Greg in Mississippi
-
Hi Greg,
How are you keeping?
Use of the SE1 with inefficient speakers will depend on how loud you tend to play your music. The CIRC1 may be the better option as it can be engineered for much greater output without too much thermal management penalty.
I have had a number of enquiries about applying these output stages, and providing voltage gain stages for them, so they are more user-friendly with normal systems. So I am going to re-open the case on these projects and also the open baffle line source array I use in my system as people have been asking about this too. It will have to be done in sections, as time allows, as I am very busy at present and have little free time. I will set up threads in the Paul Hynes Design Circle so all the discussions can be centralised and easier to find.
Regards
Paul
-
Paul,
Doing pretty well here, working too much to do much audio work right now tho!
I'm not a head-banger & am very intrigued by the Circlotron version too, but I'm slowly getting my turntable setup running again too and all of my current & potential phono stages are SE, so the SE-1 looks like a better compromise option.
I'll watch for your future posts in your circle. Keep up the great work!
Later!
Greg