Anand
Well weirdly I made and still own a pair Geddes Abbeys! The M2s are something else though and boy are they singing with the P86s driving the compression drivers. ( One of these days I'll get organised and put the Abbeys up for sale. They are great speakers, it's just that the M2s are something rather special.)
Well weirdly I made and still own a pair Geddes Abbeys! The M2s are something else though and boy are they singing with the P86s driving the compression drivers. ( One of these days I'll get organised and put the Abbeys up for sale. They are great speakers, it's just that the M2s are something rather special.)
Studley,
I owned the Abbeys previously as well. The NA12's are much more refined, detailed, etc...compared to the Abbeys, but still passive in design. The NA12's are an evolutionary step forward compared to the Abbey although the Abbey was revolutionary for me since it was my first CD design. The NS15 is fully active but unobtanium. So I'm left with the JBL 4367 (passive) & M2 (active).
I am the only owner of the NA12. I've listened to every Geddes design from the original Summa and forward, except the NS15.
We should compare notes...but via pm, of course.
Best,
Anand.
I owned the Abbeys previously as well. The NA12's are much more refined, detailed, etc...compared to the Abbeys, but still passive in design. The NA12's are an evolutionary step forward compared to the Abbey although the Abbey was revolutionary for me since it was my first CD design. The NS15 is fully active but unobtanium. So I'm left with the JBL 4367 (passive) & M2 (active).
I am the only owner of the NA12. I've listened to every Geddes design from the original Summa and forward, except the NS15.
We should compare notes...but via pm, of course.
Best,
Anand.
So your goal is to add a MOD286 stereo amp within the same chassis as the PAR86 by basically adding a heat sink and connecting the MOD286 board to Power-86 supply?
That's possible, but with the ±35 V supply, you won't get optimal performance in 4 Ω. You can expect about 55-60 W before the LM3886 starts to trip the thermal protection. This is an issue with the LM3886 itself. If you're willing to live with this limitation, you can just add the MOD286 and heat sink to your existing supply. It's just not the recommended engineering solution, that's all.
The other consideration is the amount of power the power transformer can deliver. You'll need 360-400 VA to power a 4-channel amp, assuming all channels are driven to clipping with music signal at the same time. That's a pretty unlikely scenario, so you could choose to skimp there. Considering that most listening occurs at no more than a few watt, you're probably OK. Just not a good engineering solution.
I generally avoid half-baked solutions, but it's your amp so it's ultimately your choice.
Tom
A Question for Tom
When spec'ing the Power86 board, for use driving a monoblock configuration (Modulus-86) what is your thinking regarding the amount of capacitance used for the filter capacitors?
The build documentation suggests 22,000 uF x2 63V.
I am using a 200VA transformer for each mono amplifier.
Do you feel there would be detrimental performance if one were to build the board, for the above configuration, with capacitors rated for less than 22,000 uF and if not, where would you draw the line (e.g. how much smaller would still offer good performance?).
When spec'ing the Power86 board, for use driving a monoblock configuration (Modulus-86) what is your thinking regarding the amount of capacitance used for the filter capacitors?
The build documentation suggests 22,000 uF x2 63V.
I am using a 200VA transformer for each mono amplifier.
Do you feel there would be detrimental performance if one were to build the board, for the above configuration, with capacitors rated for less than 22,000 uF and if not, where would you draw the line (e.g. how much smaller would still offer good performance?).
There are ever so many ways to figure that. And I have to mention that more transformer amperage (it is a faster charger) reduces needed capacitance. That was a really abrupt summary of an orthodox answer.
I'd like to field this question by a really weird alternative way.
As you would need 6800u for the output cap of a single rail amplifier for an ordinary 8 ohm speaker, or a wee bit over 12,000uF per rail for a split rail amplifier without an output cap. . . Then see forth to double those figures for two 8 ohm speakers, or quadruple those figures for two 4 ohm speakers.
With that latter bit, it wraps around to the orthodox answer in that quality with such a load is more reliant on the higher current transformer so as to get whatever caps you have, charged completely full, in a timespan that is shorter than the span in-between bass beats.
It has also become necessary to mention that the audio quality of caps during charging is quite useless. The full charge must be recouped directly after a bass beat in order to have clear vocals and treble (after the draining bass event). Adding a huge amount more capacitance will take considerably longer to charge full and thus reach quality. So, adding excess capacitance will not serve. Of course more capacitance could be made up for with a much heavier transformer, capable of higher current, aka faster charge.
With these matters in mind, which is certainly enough to have me pacing the floor for quite a while, I think it would be good to firstly ask how much transformer for the given load, and then afterwards ask how much can one further increase the load by adding capacitors that one might not need.
Tom's proposed power supply is more than enough for 2 of 8 ohm speakers. I have used 13,200uF per rail as a minimum, the average figure is 15,000uF on orthodox measure at this average wattage range, and if you have 22,000uF, that'll suit even a pair of 6 ohm speakers.
If you have heavier loads than that, for the parallel amplifier, then:
pair 3 ohm loads, the parallel amplifier, voltage same as suggested for mod86, and double the transformer current figure.
pair 4 ohm loads, the parallel amplifier, and you'd be welcome to try more capacitance, if prior to that you'd increased available transformer current by same proportion. Well, it is a linear amp.
As for the orthodox answer versus the weird alternative, you actually have to satisfy both requirements if you want an ordinary amplifier to work right. However, as Tom's products have a stabilized front end, they're quite likely to perform just as nicely even if you've goofed the capacitance and current figures at the power supply.
I have made the above post without regard to the immunity of Neurochome amplifiers to such factors. I hold true that decreasing the need of negative feedback also decreases the sound of negative feedback. In other words, if you don't overburden the excellent technology, then it may work even better. Well, that's the theory on it.
On the first glance I missed a bit. Okay, that plan will work (except voltage?). However, if you have 4 ohm speakers, then 300va or even 400va transformers could be used.
When you'd need to up the transformer is also relevant to the size space to fill with sound. The larger transformers can serve in the country club, or an exceptionally large size house with an open floor plan.
In my experience the 200va monoblocks work fantastically in normal size quarters. However, if the space is much bigger, a similar effectiveness takes more current. I would experience some cowardice if the was a plan to use 400va monoblocks of either Neurochrome design or mine within a room that wasn't easily considered gigantic. There's some risk to ear health with too much power in a too small area.
Please proceed with the details on the speaker load and room size.
Edit: Also, that transformer voltage is high, as in a bit unorthodox. I would have expected 25+25vac to 28+28vac for the parallel amplifier. It is true that 30+30vac transformer has been used without explosion. However, your quote of 63+63vac does not sound useful to me, because I think it will make the amplifiers either explode or turn off, probably turn off before they've reached potential. I've got to mention that too much voltage will reduce available power output considerably. Might want to check that.
Last edited:
My thought was to use Par86 for woofers & 286 for tweeters. but now it seems I should get a different transformer or add a second transformer and use the AS3224 for the Par86. Also seems I'd need a separate volume control for either the par86 or the 286....not neat and hard to balance sound. I was just looking for a cheap way to bi-amp using what I already have by adding minimal parts.
Nolonger the plan. I'm Skipping bi-amping for now.Thanks for the info.
danielwritesbac:
The transformer is 200 VA 22V; 63V isn't referring to the transformer output or the DC rail voltage, it's the capacitor's rated voltage.
The Power86 documentation calls for 2x 22,000 uF 50V, but Mouser didn't have the 50V from Tom's project number in stock when I built my BOM, thus 63V types under consideration.
Now, the Power86 supply is designed to run a stereo configuration, while my application is a single channel. However, if the capacitance is there to take care of ripple current, which would be high(er) with a larger VA rated transformer, that doesn't offer an answer with regard to using a lower uF value one way or the other.
On the other hand, the options for smaller capacitance snap-ins w/10mm lead spacing are more numerous and are available in a greater number of series.
Rod Elliot [see link] suggests much lower values should be adequate for a Class AB configuration; his suggested minimum is 2~3000 uF per volt output into 8 ohms, depending on line voltage (220/50 or 120/60). Going by his way of thinking 10,000 uF should be more than adequate.
Thus my question, directed at Tom because, after all, he designed the Power86 board and the Modulus86, so he must have his reasons for the suggested capacitance, which I am curious to understand.
The loudspeaker application doesn't present problems in my opinion; the amp will be used with Energy Connisseur C-2's which dip near but not quite 4 ohms over part of the audioband, but are more than adequate SPL-wise in the area3 of my listening room.
Currently powered by a 40wpc EL34-based Anthem Amp-1, which doesn't reach the stress point to my ears, at least. Naturally that isn't equal to 40 solid state watts but with the impedance I think the Modulus86 should be able to squeeze a bit more, mitigating somewhat the difference in overload characteristics between Vacuum State and Solid State.
However I do want the Modulus86 to be able to drive a 5 ohm load without undue stress. Configuration is with that in mind.
I have read some papers and essays on the subject, aside from Rod Elliot [ESP], application notes from Linear Technology and a few others ("The Importance of the Power Supply", 1997, Nelson Pass).
Linear Power Supply Design
The transformer is 200 VA 22V; 63V isn't referring to the transformer output or the DC rail voltage, it's the capacitor's rated voltage.
The Power86 documentation calls for 2x 22,000 uF 50V, but Mouser didn't have the 50V from Tom's project number in stock when I built my BOM, thus 63V types under consideration.
Now, the Power86 supply is designed to run a stereo configuration, while my application is a single channel. However, if the capacitance is there to take care of ripple current, which would be high(er) with a larger VA rated transformer, that doesn't offer an answer with regard to using a lower uF value one way or the other.
On the other hand, the options for smaller capacitance snap-ins w/10mm lead spacing are more numerous and are available in a greater number of series.
Rod Elliot [see link] suggests much lower values should be adequate for a Class AB configuration; his suggested minimum is 2~3000 uF per volt output into 8 ohms, depending on line voltage (220/50 or 120/60). Going by his way of thinking 10,000 uF should be more than adequate.
Thus my question, directed at Tom because, after all, he designed the Power86 board and the Modulus86, so he must have his reasons for the suggested capacitance, which I am curious to understand.
The loudspeaker application doesn't present problems in my opinion; the amp will be used with Energy Connisseur C-2's which dip near but not quite 4 ohms over part of the audioband, but are more than adequate SPL-wise in the area3 of my listening room.
Currently powered by a 40wpc EL34-based Anthem Amp-1, which doesn't reach the stress point to my ears, at least. Naturally that isn't equal to 40 solid state watts but with the impedance I think the Modulus86 should be able to squeeze a bit more, mitigating somewhat the difference in overload characteristics between Vacuum State and Solid State.
However I do want the Modulus86 to be able to drive a 5 ohm load without undue stress. Configuration is with that in mind.
I have read some papers and essays on the subject, aside from Rod Elliot [ESP], application notes from Linear Technology and a few others ("The Importance of the Power Supply", 1997, Nelson Pass).
Linear Power Supply Design
Last edited:
I see where you're coming from. Sadly, I think you've reached the wrong conclusion. The ampacity of a transformer is determined by the core area and the diameter of the wire used for the primary and secondary. A higher ampacity transformer will be able to deliver higher current without melting the insulation on the wires. It will also have lower DCR of the windings. However, a transformer is a voltage source; not a current source. Thus, switching from a 100 VA transformer to a 1 kVA transformer will do very, very little for the ripple voltage on the output of the power supply. The very, very little it will do is due to the change in DCR, which frankly is negligible in this context.
Loading an unregulated power supply, such as the Power-86, will result in ripple on the output voltage. The lower the supply capacitance, the higher the ripple. You can estimate the amount of ripple from the definition of capacitance:
C = Q/V, where Q is the charge stored in the cap, V is the voltage across the cap, and C is the capacitance.
Furthermore, i = Q/t --> C = I*t/V <--> V = I*t/C.
t is the time between charge pulses, i.e. 1/(2*f_mains), where f_mains is the mains frequency.
It should be fairly obvious from the equation above that by increasing I (increase output current) or decreasing C (supply capacitance), the ripple voltage, V, will rise. Always.
The above math is an estimate as it assumes zero conduction angle on the rectifier diodes. That's obviously not the case in practice. For a better estimate of the supply ripple, I suggest using a SPICE simulator or a power supply design tool such as PSUD II.
The ripple voltage causes the supply voltage to droop under heavy load. Thus, the max amount of tolerable ripple voltage is determined by how much you want the supply voltage to droop under load. This droop eats directly into your max output power spec.
The other consideration is that for most amplifiers, some portion of supply ripple will make it through to the amp output, thus degrade sound quality by causing IMD products. This is not the case for the Modulus-series of amps due to their incredibly high power supply rejection so you have a little bit more margin for supply droop there.
Personally, I'd say 10000 uF would be fine for a mono amp. I'd go with 22000 uF for a stereo build. Capacitors like that are really quite inexpensive these days. Why skimp?
Tom
63 V caps will be fine. With that transformer, you could go with 40 V types too. As long as they're the snap-in type with 10 mm pin pitch and a specified diameter of 40 mm or less, they'll work fine in the Power-86.
Tom
Thanks, Tom. Very helpful.
It's not so much a question of cost as one of availability. I have no problem ponying up for 22,000 uF types, but am seeking guidance (which you have provided) on my options. Some flexibility in overall capacitance means more choice of manufacturer and type. Mouser offered me an "in stock" choice of ... one.
It's not so much a question of cost as one of availability. I have no problem ponying up for 22,000 uF types, but am seeking guidance (which you have provided) on my options. Some flexibility in overall capacitance means more choice of manufacturer and type. Mouser offered me an "in stock" choice of ... one.
Last edited:
While we're talking power supplies... I have on my want-to-do list a compact-as-possible Modulus-286 build. The compact goal favors a switch-mode power supply.
I know of the Connex SMPS units, and also Tom's own SMPSes. But in the interest of being cheap... both Ebay and AliExpress have these cheap high-power dual-rail switch mode PSUs. Ebay listing: 500W Amplifier Switching Power Supply Board Dual-voltage PSU +/-55V. AliExpress listing: 500W amplifier switching power supply board dual-voltage PSU +/-55V. While the headline lists +/-55V, they can be configured down to +/-30V.
(I understand that 30V supply could hit the lm3886's current limit with 4ohm loads. But at my modest listening levels, I have no concerns of getting anywhere near any power limits.)
I guess my question is: given the Mod[2]86's excellent PSRR, is it OK to go with a super-cheap (possibly sketchy) power supply?
I know of the Connex SMPS units, and also Tom's own SMPSes. But in the interest of being cheap... both Ebay and AliExpress have these cheap high-power dual-rail switch mode PSUs. Ebay listing: 500W Amplifier Switching Power Supply Board Dual-voltage PSU +/-55V. AliExpress listing: 500W amplifier switching power supply board dual-voltage PSU +/-55V. While the headline lists +/-55V, they can be configured down to +/-30V.
(I understand that 30V supply could hit the lm3886's current limit with 4ohm loads. But at my modest listening levels, I have no concerns of getting anywhere near any power limits.)
I guess my question is: given the Mod[2]86's excellent PSRR, is it OK to go with a super-cheap (possibly sketchy) power supply?
This is nonsense. The cap on the output of a single rail amplifier acts as a high pass filter and will limit low frequency performance if it is too small. The filter caps in a power supply are filtering 100/120Hz. If the amplifier attached to the power supply has good PSRR (Like the subject amplifiers), the only impact of having too small filter caps is a reduction of maximum power output before clipping, it has no impact on the frequency response of the amplifier.
I understand you are trying to be helpful Daniel, but in many cases you are doing little more than adding confusion to what are very simple concepts.
For a stereo MOD286, the Connex SMPS300RE (±30 V version) is perfect. For a 4-channel amp, I'd go with the SMPS500R (±30 V version).
Or to paraphrase: Can I get my 2018 Audi RS8 configured with $50 Walmart tires, please. I'm trying to save a buck.
The quality of Alibaba/eBay/etc. electronics is dubious to say the least. Sometimes you get what you pay for. Many times you don't. If the supply fails catastrophically, it might take both the MOD286 and the speakers with it.
The Connex supplies tend to fail to 0 V when they die due to overload (ask me how I know). They just stop putting out voltage. That's the best scenario for a supply failure.
The majority of the cost will be the chassis, board, and parts. A Connex supply is maybe 15-20 % of the total build cost.
Tom
This may not be of use to everyone here, but I just encountered an interesting thing doing a part search.
The usual suppliers ... DigiKey, Mouser, etc ... have different sites for US and Canada. DigiKey in particular always makes clear that on the Canadian site, prices are in $C and include duties, taxes, brokerage and any other fees, which are handled by DigiKey themselves. I naturally assumed Mouser, etc operated in the same way, and have never encountered an example where that was not the case. Knowing the above, I had always assumed that parts were sourced from the US warehouse, which in the case of DigiKey, is not far from me, as it turns out, in Thief River Falls, MN.*
So ... I'm looking for a certain capacitor. I do a search and arrive at the US site of DigiKey, where they indicate it is out of stock and is expected to be restocked sometime in July.
Wanting the $C price, to do some evaluations, I entered the proper URL and copied and pasted the part #. Well, lo and behold, there are some 600 in stock.
So, lesson learned, mostly of use to people sourcing parts from Canada, of course. Don't assume your Google search, which may send you to the US site when doing parts research, and especially if you are using links provided in forums such as DIYAudio, where people tend to provide links to US sources, is the whole story. Copy your parts numbers or enter your BOMs and Project Numbers in the appropriate site. It might be a mistake to assume parts will be sourced from the main warehouse in America and processed through the corporate brokerage for your all-in end cost. Maybe the parts are sitting locally, ready for immediate shipment.
* I had the opportunity to spend a few days with the founder of DigiKey many years ago, on a fishing trip. Great guy, and a great story about how they came to be a major parts supplier from some rather humble beginnings, "in the middle of nowhere".
The usual suppliers ... DigiKey, Mouser, etc ... have different sites for US and Canada. DigiKey in particular always makes clear that on the Canadian site, prices are in $C and include duties, taxes, brokerage and any other fees, which are handled by DigiKey themselves. I naturally assumed Mouser, etc operated in the same way, and have never encountered an example where that was not the case. Knowing the above, I had always assumed that parts were sourced from the US warehouse, which in the case of DigiKey, is not far from me, as it turns out, in Thief River Falls, MN.*
So ... I'm looking for a certain capacitor. I do a search and arrive at the US site of DigiKey, where they indicate it is out of stock and is expected to be restocked sometime in July.
Wanting the $C price, to do some evaluations, I entered the proper URL and copied and pasted the part #. Well, lo and behold, there are some 600 in stock.
So, lesson learned, mostly of use to people sourcing parts from Canada, of course. Don't assume your Google search, which may send you to the US site when doing parts research, and especially if you are using links provided in forums such as DIYAudio, where people tend to provide links to US sources, is the whole story. Copy your parts numbers or enter your BOMs and Project Numbers in the appropriate site. It might be a mistake to assume parts will be sourced from the main warehouse in America and processed through the corporate brokerage for your all-in end cost. Maybe the parts are sitting locally, ready for immediate shipment.
* I had the opportunity to spend a few days with the founder of DigiKey many years ago, on a fishing trip. Great guy, and a great story about how they came to be a major parts supplier from some rather humble beginnings, "in the middle of nowhere".
Last edited:
I rarely type actual part numbers into common search engines. I go directly to Mouser/Digikey/Newark/etc. That's always given me the correct pricing and availability. Thanks for the warning, though.
Digikey is awesome. I wish more DIYers used them. Every order I place through Digikey gets here overnight, whereas Mouser sits on the box for a day as they type in the customs paperwork. Both are good suppliers, Digikey is just faster.
I do think Mouser's way of doing projects is better thought out. Digikey is catching up, though. Both support BOM upload to cart now.
Tom
Digikey is awesome. I wish more DIYers used them. Every order I place through Digikey gets here overnight, whereas Mouser sits on the box for a day as they type in the customs paperwork. Both are good suppliers, Digikey is just faster.
I do think Mouser's way of doing projects is better thought out. Digikey is catching up, though. Both support BOM upload to cart now.
Tom
Completely agreed.
And yet there is this pervasive notion of using "small" valued filter capacitors because it makes the amp sound audibly "faster" etc...I've seen this mentioned numerous times in the Chipamps forum and on the AudioSector subforums. Mind you those designs are a FAR FAR cry from even an LM3886DR design so perhaps the incompetencies of those designs are really what the subjective appraisals are all about.
I myself am guilty of falling down the rabbit hole that all LM3886 designs are similar/same...never again!
Tom's post regarding reviewing the definition of capacitance and how it relates to ripple voltage and charge stored along with an understanding of programs like PSUD II is all one really needs.
No need to reinvent the wheel eh?
Best,
Anand.
I think that was about 2900 posts ago. Basically no difference other than power output between a lab supply, a simple unsnubbed transformer, rectifier and caps and SMPS. PSRR does count for something.
Power output differences were mainly due to differing rail voltages and current capabilities IIRC.
Power output differences were mainly due to differing rail voltages and current capabilities IIRC.