UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 3rd Dec 2018, 5:27 pm   #1
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Reservoir capacitors and ripple current.

If this isn't the relevant discussion forum mods please move it.
Ok, I'm modelling the PSU for a stereo power amp build.
It's for 4 x EL34 with a total current draw around the 300mA mark, this includes the calculated draw for preamp and phase splitter valves.
So far so good, the PSU modelling software gives me 450V across the first reservoir capacitor. Using a standard silicon bridge rectifier rated at 25amps 1kV.
After that it's choke and capacitor smoothed to the CT of the output transformers.
So I was looking at the choices of reservoir capacitor with a high enough working voltage and 105 degC max temp and for pennies more I can get a 470 uF capacitor with a ripple current of 2.15A which seems pretty good.
but I think the ripple current could exceed this value? Am i right in thinking that a higher reservoir capacitance comes at the cost of higher ripple current as the charging pulses will be very high in value?
So what, if any, benefit do I get by increasing reservoir capacitance above say 16uF where the psu model shows a more modest ripple current of about 1.5A as I think if I use anything above 100uF the ripple current goes up to 2,5A or have I made an error in the calculation/interpretation of the simulation software?
There shouldn't be any noticeable hum, the choke and subsequent capacitor and resistance networks to each stage should filter that adequately and I'd expect a reasonably balanced PP output to cancel a lot of PSU hum out.
The transformer is massively over specified and would supply 8 x EL34 if asked so I assume it will be capable on its own of providing a bit extra current on peaks?
I will have made a number of schoolboy errors here I know. I could just run with what I know works but I'm trying to get an understanding of the way a PSU does its thing as part of this DIY project.
Sorry its a long question.
Andy.
bikerhifinut is offline  
Old 3rd Dec 2018, 6:10 pm   #2
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: reservoir capacitors and ripple current

Quote:
Am i right in thinking that a higher reservoir capacitance comes at the cost of higher ripple current as the charging pulses will be very high in value?
So what, if any, benefit do I get by increasing reservoir capacitance above say 16uF where the psu model shows a more modest ripple current of about 1.5A as I think if I use anything above 100uF the ripple current goes up to 2,5A or have I made an error in the calculation/interpretation of the simulation software?
Adding more capacitance will always increase ripple current.

However, it could well be that the things which determine ripple current are already dominated by series resistance (such as transformer winding resistance), in which case increasing capacitance will hardly make much difference. Modelling will show you if this is true.

More capacitance will give you less ripple voltage. You can make this as low as you like by increasing capacitance sufficiently.

More capacitance also means a bigger can, so more surface area to radiate heat and keep cool.

More capacitance (mainly smoothing rather than reservoir) also means the HT rail is more 'solid' at low frequencies - you are less likely to have trouble with motorboating.

However, more capacitance means the thing takes longer to charge. If you use a kilofarad's worth of capacitance, you will blow your mains fuses every time you try to switch on. This will be even worse if you have a massively over-specified mains transformer, with really low winding resistances. Sometimes bigger ain't necessarily better!
kalee20 is offline  
Old 3rd Dec 2018, 6:33 pm   #3
turretslug
Dekatron
 
turretslug's Avatar
 
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,385
Default Re: Reservoir capacitors and ripple current.

Further to the surface area point, it's not so unusual to find two or more parallelled reservoir capacitors in place of one bigger one where ripple rating looks to be marginal or doubtful- say 2x 220uF or even 3x 150uF in place of the 470uF will nearly always have a usefully increased combined ripple rating for capacitors from the same series etc. Even in a high-power valve amp, you won't run into the need for chunky capacitor-linking busbars that a PSU of the same wattage but one-tenth the voltage might need to properly share the ripple.
turretslug is offline  
Old 3rd Dec 2018, 7:45 pm   #4
Mooly
Octode
 
Join Date: May 2011
Location: Lancashire, UK.
Posts: 1,349
Default Re: Reservoir capacitors and ripple current.

Have a look at this. The single 470uF cap has a series resistance of 0.12 ohms which seemed a realistic value. The voltage source (transformer) has four different series 'resistance' values set, 0.1 ohm, 1 ohm, 2 ohm and 5 ohm. This series 'resistance' makes a massive difference to the ripple current seen in the cap. You can of course add your own series resistance to bring down the peak ripple currents seen.

The first image shows the ripple current in the cap, the second image shows the DC supply and corresponding ripple for these different series impedances.
Attached Thumbnails
Click image for larger version

Name:	Annotation 2018-12-03 183849.jpg
Views:	163
Size:	79.3 KB
ID:	173883   Click image for larger version

Name:	Annotation 2018-12-03 183936.jpg
Views:	150
Size:	81.5 KB
ID:	173884  
Mooly is offline  
Old 3rd Dec 2018, 10:44 pm   #5
Ed_Dinning
Dekatron
 
Join Date: Jan 2004
Location: Newcastle upon Tyne, Tyne & Wear, UK.
Posts: 8,172
Default Re: Reservoir capacitors and ripple current.

Hi Andy, have a good dig on the manufacturers website. They normally publish tables of ripple current , ESR, lifespan and temp rating for each of their cap ranges.
We used this a lot in SMPS design.
Note that the same cap in different aspect ratios can have different max current and temp ratings.

I have some of the charts here but they are for older caps, no longer current.
Reduction of cap temp by 10C will double lifespan. Figures quoted on these charts are often for 50 or 100KHZ ripple. 50/100Hz is normally much more benign.

Ed
Ed_Dinning is online now  
Old 3rd Dec 2018, 10:52 pm   #6
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Thanks for the responses, much appreciated.

I did a remodel on the Duncans amps PSU2 software.

I know the transformer has a DC R of 7 ohms so i use that, likewise the chokes are 100 ohms each and effectively would be in parallel as they'd split the supply after the first reservoir capacitor downstream of the rectifier. So I assumed a DC R of 50 ohms for the chokes.
The capacitor data gives 200 milliohms ESR.
I reduced the capacitance of the first reservoir to 100uF and made the capacitance after the chokes up to 470uF each side, I also ran it as a parallel circuit assuming 1000uF after the choke which would be 2.5H at 50 ohms DC R.
The simulation showed the ripple current across C1 ie the reservoir to drop to around 500mA. I assume the resustance of the chokes slows down the inrush/charging current to the big capacitors and the ripple is reduced sufficiently to not cause any big effects.

But the question is, does this present any advantage on High transient peaks or does the presence of the resistance of the chokes slow down recharging time sufficiently to cause problems when the amp is responding to high peak levels?
I think there would be no audible and little measurable advantage in terms of hum as a simple 50uF-5H -50uF filter renders the amps virtually silent as is.
I am interested in the findings that paralleling capacitors increases their capabilities of handling ripple and as I already have plenty of high quality F&T 50+50 uF capacitors with a very high ripple rating, I am tempted to make a single 200 uF reservoir from 2 of these as I have the holes already made in the top panel. The question then is, do I need to have at least 100uF on each stereo channel after their respective chokes, or would 50uF be enough as part of the C L C filter? post this inital PSU filter the capacitors form part of a series of RC filters formed from voltage dropping resistors and their associated bypass capacitors and their values are not so critical although I will likely keep them at about 50uF as that is the values I have at 500V DC Wkg. I could argue that the HT voltages are at about 350 and 150 V respectively but my feelings are using silicon rectifiers in particular means the full HT could be applied across all PSU capacitors until the valves start to conduct and the voltage through the resistors drops.
Thanks for the warning about initial inrush current into very high reservoirs too, does this mean a trial and error on mains fuses until I get a value that doesnt blow on switch on? Normally i would calculate a rough current draw based on the amplifiers power needs and use that as a starting point for slow blow (T) fuses and use the smallest value available that didnt blow on switch on.
On that subject, I have seen designs that put a resistance in series with each diode of around 10 ohms, I suppose thats a form of inrush current limiter? I'd only lose about 1.5V through each diode so thats a bearable trade off. I'll also be putting a high voltage 10nF capacitor across each rectifier diode to suppress switching spikes.

Many questions again sorry, as I said before I am attempting to gain a better understanding of what is actually happening in the circuit rather than just following a "recipe".

Andy.
bikerhifinut is offline  
Old 3rd Dec 2018, 11:00 pm   #7
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Quote:
Originally Posted by Ed_Dinning View Post
Hi Andy, have a good dig on the manufacturers website. They normally publish tables of ripple current , ESR, lifespan and temp rating for each of their cap ranges.
We used this a lot in SMPS design.
Note that the same cap in different aspect ratios can have different max current and temp ratings.

I have some of the charts here but they are for older caps, no longer current.
Reduction of cap temp by 10C will double lifespan. Figures quoted on these charts are often for 50 or 100KHZ ripple. 50/100Hz is normally much more benign.
Cheers Ed, that is exactly where I started with all this by downloading the PDF files of the makers data for the capacitors I was interested in. And I did find the specs for 50 and 60 Hz. I already knew about the increase in lifespan by keeping em cool. I tend to try and buy 105 degC caps on that assumption. It was interesting to see the phenomenal increase in life expectancy with reduced temps. The Vishay data showed a lifespan of 700,000 hours at 40 C but down to 3000 hours at 85 which was quite sobering and a good incentive to make the effort to mount the capacitors above decks so to speak in the air.
If I learn something as well as getting the satisfaction of saying "I built that from scratch" it has to be a win win?
I surprised myself when I managed to rewind/alter a heater secondary on a toroid. Admittedly its probably a doddle to you transformer experts but it was a huge leap of faith for me!

A.

Last edited by bikerhifinut; 3rd Dec 2018 at 11:02 pm. Reason: capitalisation and punktuashun.
bikerhifinut is offline  
Old 4th Dec 2018, 12:23 pm   #8
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: Reservoir capacitors and ripple current.

Quote:
Originally Posted by bikerhifinut View Post
I assumed a DC R of 50 ohms for the chokes.

The capacitor data gives 200 milliohms ESR.

I also ran it as a parallel circuit assuming 1000uF after the choke which would be 2.5H at 50 ohms DC R.

I assume the resustance of the chokes slows down the inrush/charging current to the big capacitors and the ripple is reduced sufficiently to not cause any big effects.

But the question is, does this present any advantage on High transient peaks or does the presence of the resistance of the chokes slow down recharging time sufficiently to cause problems when the amp is responding to high peak levels?
It's worth doing some basic calculations on the LC smoothing filter. You say 2.5H and 1,000μF, well that gives a resonant frequency of 3.2Hz. (So it will do a really good job of filtering any 100Hz ripple on the reservoir - a reduction of over 900x in fact).

The Q of this filter will be determined by the choke resistance (you say it is 50 ohm), the ESR of the capacitor (0.2 ohm so forget it) and the damping provided by the amplifier (which we don't know). With just the choke resistance, I make it a Q of 1. So the thing will not be prone to a lot of ringing and you will not have to take precautions to avoid oscillations at 3.2Hz. From this point of view, the choke's resistance is A Good Thing.

After a sudden transient demand, you can get a feel for the recharge time - even if the reservoir voltage stayed absolutely constant, the smoothing capacitor will take (very crudely) 1/2 a period to recharge, ie 157msec. Since the Q is 1, the recharge time is determined roughly equally by the DC resistance and by the inductance.

If you increase the smoothing capacitor further, the time will be even longer - but of course with a larger value, the transient demand will have resulted in a smaller voltage drop to recharge from. So recharging time is not the whole story!

Last edited by kalee20; 4th Dec 2018 at 12:25 pm. Reason: Clarity & typos
kalee20 is offline  
Old 4th Dec 2018, 12:55 pm   #9
Craig Sawyers
Dekatron
 
Craig Sawyers's Avatar
 
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,941
Default Re: Reservoir capacitors and ripple current.

LC smoothing used to be a favourite method of reducing ripple in the days when wound components were not ridiculously expensive. As kalee says if the Q at the resonant frequency is high, there is a tendency to ring particularly if there is no load. LC power supplies in which the load was either there or not or varied widely, there was usually a power resistor to draw sufficient current under all load conditions in order to suppress ringing.
Craig Sawyers is offline  
Old 4th Dec 2018, 5:08 pm   #10
Guest
Guest
 
Posts: n/a
Default Re: Reservoir capacitors and ripple current.

Quote:
when wound components were not ridiculously expensive
They where, as where all the other bits, a case of balanced expenditure.
 
Old 5th Dec 2018, 1:36 am   #11
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Grand stuff gents,
talking about ringing thats what my ears are doing after being in Carlisle tonight at a Pink Floyd tribute band gig with my sis for her 60th, loud? Viscerally so!


Anyway the amp design I have roughed out will be mostly biased into ClassA with only hopefully odd excursions into class AB which as far as I understand it means the current draw is as a rule constant?
I've kind of come to a conclusion which is to use a bank of 50+50 high ripple caps to make a single 200uF reservoir which feeds two separate 5H by 100 ohm chokes and thence to a 100uF smoother followed by the output transformer CT, the following 2 stages will be simply decoupled with 50uF wire ended jobs under the hood as it were. These caps feeding low current draw stages should stay cool so I feel confident.
I suspect huge reservoir and smoothing caps nay have more of a "brag" value than any really useful contribution to overall amplifier performance. And best of all, I have plenty in the toybox so won't need to buy them off RS!
I think thats the best compromise overall. My PSU simulator doesnt seem to work out the resonant frequency of the filter so I'll need to bone up on the maths.
Thanks for the enlightenment gents.
This amp may yet fly in time for Christmas.

A.
bikerhifinut is offline  
Old 5th Dec 2018, 1:38 am   #12
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Quote:
Originally Posted by merlinmaxwell View Post
Quote:
when wound components were not ridiculously expensive
They where, as where all the other bits, a case of balanced expenditure.
At a DIY level even high current chokes aren't terribly expensive and mostly for one offs we don't need to substitute in a 10p resistor in order to squeeze a bit of profit!
I think thats what you meant merlin?
A.
bikerhifinut is offline  
Old 5th Dec 2018, 9:42 am   #13
wd40addict
Octode
 
wd40addict's Avatar
 
Join Date: May 2009
Location: Stevenage, Herts. UK.
Posts: 1,515
Default Re: Reservoir capacitors and ripple current.

You need to use LTSpice RMS function on your current simulation. Capacitor ripple currents are normally quoted RMS. Your results are spiky, but probably well below 2.15A!

https://www.analog.com/en/technical-...f-a-trace.html
wd40addict is offline  
Old 5th Dec 2018, 11:17 am   #14
Mooly
Octode
 
Join Date: May 2011
Location: Lancashire, UK.
Posts: 1,349
Default Re: Reservoir capacitors and ripple current.

Who me

You can't run averaging on stepped simulations but this shows a worst case scenario with a 0.1 ohm impedance of the voltage source. 1.4A RMS
Attached Thumbnails
Click image for larger version

Name:	Untitled.jpg
Views:	69
Size:	94.0 KB
ID:	173986  
Mooly is offline  
Old 5th Dec 2018, 11:31 am   #15
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Quote:
Originally Posted by wd40addict View Post
You need to use LTSpice RMS function on your current simulation. Capacitor ripple currents are normally quoted RMS. Your results are spiky, but probably well below 2.15A!

https://www.analog.com/en/technical-...f-a-trace.html
ah!
I was taking the maximum current as the limiting factor, the RMS simulates at about 750mA which is well inside the 2.15A rating of the capacitors I was considering.
If that's the case then I'm ok with the bigger values.
So.............
Whats the wisdom on this? I am thinking that a big 470uF reservoir after the rectifiers and then split into two rails via the 5H chokes and only 50uF after the choke to the CT of OPT txfrmer. That should cover all eventualities?
bikerhifinut is offline  
Old 6th Dec 2018, 7:03 pm   #16
PJL
Dekatron
 
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
Default Re: Reservoir capacitors and ripple current.

I find it difficult to believe the HT secondary is just 7 ohms? That would be a very large transformer which raises the observation that over specifying one component does mean you end up having to use better parts throughout. I assume this is a CLC design?
PJL is offline  
Old 6th Dec 2018, 11:13 pm   #17
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Thats what I measured across the HT secondary with both analogue and digital meters and I assumed that was the practical DC source resistance for the simulator.
I covered the mains transformer in an associated post, I managed to alter the turns ratio on one of the heater secondaries from 9V to 6.3V for the 4 EL34 output valves.
It's a hand wound one obtained from another forum member and was originally intended in a previous life to power up a monster amp using 4 x EL34 or similar per channel so its cruising along at something less than half capacity. It's not really a question of over specifying so much as making use of what's available.
Anyway it is what it is.
The plan is to more or less stick with the original circuit which as you divined is a CLC to the output transformer CT, followed by a couple of RC filters to the phase splitter and first stage valves. It's just that now I am using silicon to rectify the HT instead of a GZ34 I can afford the luxury of a larger value reservoir capacitor over the original maximum 50uF value. The cost of a single good quality 470uF capacitor isn't very much more than a 50 or 100 uF which is where I came in with the queries. It's quite possible that sticking with 100uF as the reservoir capacitor will be more than plenty and certainly I have plenty of those rattling around in the toybox.
But there seems to be a school of thought that a big reservoir capacitor can help on occasional large transients. My initial concern was more about the time to recharge the bigger capacitor and any ill effects that may have. Then I got involved in the ripple calculation as the roughed out circuit progressed.
Ok for further info, after the first reservoir capacitor the HT will be split to left and right via a 5H choke and the output end of those will each have a capacitor to earth and that feeds each individual output transformer. I keep individual supplies from there to each stage on either side in a sort of pseudo double mono arrangement. That's because the chokes are rated at 150mA each and the amps draw 145mA each side so it's either parallel em and keep a single supply rail or split em. Probably no audible advantage but I can do it so I will.
bikerhifinut is offline  
Old 7th Dec 2018, 2:25 am   #18
joebog1
Banned
 
Join Date: Jun 2015
Location: Mareeba, North Queensland, Australia
Posts: 2,704
Default Re: Reservoir capacitors and ripple current.

Quote:
Originally Posted by PJL View Post
I find it difficult to believe the HT secondary is just 7 ohms? That would be a very large transformer which raises the observation that over specifying one component does mean you end up having to use better parts throughout. I assume this is a CLC design?
The secondary will be very close to 7 ohms!!.
I "donated" the transformer to a member that had his chassis shimmering at 50 Hz due to massive flux.
The core is a 650 watter, and then I derated it. I then used wire in the range 600-750 CM/A ( dont blow yer cool man, said the fly to mad John)

Biggest problem is inrush current at switch on, and back EMF at switch off ( unless you can "see" the peaks and troughs @50Hz ) The transformer will drive almost ANY capacitance biker can throw at it, PROVIDING he has rectifiers that can handle the inrush current. Having had the experience of "cheap" transformers that can vibrate a steel chassis similar to an industrial ultrasonic cleaner, I "donated" the above traffo.

Secondary "resistance" about 7 or 8 ohms DC
Primary "resistance around 3 ohms DC.
Weight is around 4.5 kilograms.
Temperature rise at MAXIMUM current will be less than 40C above ambient and I designed it for Northern Australia where we have just had two weeks approaching 40C.

bikerhifinut is ALSO a biker, hence the donation.

Regards, Joe.
joebog1 is offline  
Old 7th Dec 2018, 11:11 am   #19
bikerhifinut
Octode
 
Join Date: Nov 2013
Location: Penrith, Cumbria, UK
Posts: 1,993
Default Re: Reservoir capacitors and ripple current.

Cheers Joe.
I know this thing's having a longer gestation period than a Blue whale but you know the issues I had previously and the ONE thing I really have to get right is the PSU as the circuit is pretty much sorted and refined to my satisfaction.
It certainly is totally silent in operation when I ran it up on load on a test bed. And no discernible temp rise. So it's all looking good. I reckon I may as well stick with plan A which is keep the reservoir at the original design 200uF followed by 100uF south of each choke. That will work and the amp will be stable at least as far as the power supply influences things.

A.
bikerhifinut is offline  
Old 7th Dec 2018, 11:29 am   #20
joebog1
Banned
 
Join Date: Jun 2015
Location: Mareeba, North Queensland, Australia
Posts: 2,704
Default Re: Reservoir capacitors and ripple current.

As you watch the frost settle, and I wait for the wet season, BUILD the BLDY amp ,
I have pics to post of a new pre-amp/phono stage.
I promise if you do !!!

Joe
joebog1 is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 10:54 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.