Ferrite ID

A reasonably tight fit gives better results, as does using a ferrite tube rather than a short bead. However, if its a loose fit you may be able to feed multiple turns through it which works well at lower freequencies. The material does make a difference too, according to the frequency range you are trying to absorb. John

RM ferrite cores are available in a fair range of sizes.

Broad-line distributors tend to stock them, but they rarely have the whole range.

Bigger cores go up to RM14 and you might be able to pack in a couple of turns of power lead. RM4 is the smallest, and probably too small for the application.

A single turn around an RM10 core is worth about 5uH (from 1.6uhH to

If you dig into the application notes you can find out about magnetic saturation.

You were probably thinking about ferrite beads/ferrite tubes, but they tend to be a bit idiosyncratic.

Probably diode reverse recovery, although I don't see why they would care to rectify.

Is this a resistive water heater? That's very inefficient.

Tankless implies instant heating, which uses a lot of power. It is remarkably efficient, in that you don't waste any of the energy keeping the environment around storage tank warm. You'd get more heat per watt if you used a heat-pump, but you have to collect most of that heat from the environment, which is slow and call for bulky heat-exchangers.

Sharp spikes at the peaks of the positive and negative cycles doesn't necessarily imply diode rectifiers - if they just connected to the main only at the peaks of the cycles you'd get the same kind of brief current spike.

It's totally incompetent design - everybody should know how to do power factor correction - but the extra components cost money, and tankless water heaters are strictly from hunger.

First, need to define the rise time and shape of the spike, to work out how much inductance is needed to filter it, along with any caps at the supply side, for low pass filtering. You could also try separating the feed to the heater, from everything else, to reduce voltage drop effects. 40 amps is pretty high current for a domestic setup, though electric showers often match that, or more.

Sounds like a poor quality emc side of the design though.

We leave the heat on here all year. I just heard it kick on. So heat leakage from our gas water heater, and heat leaking into the house from hot-water pipes, or heat from gas cooking, are all good.

San Francisco is planning to require all new water heaters to be heat pumps, so I just had our gas heater replaced with a really good one. And I plan to replace the sacrficial anodes every few years.

I've never talked to a plumber who had heard of sacrificial anodes!

You'd get more heat per watt

A resistive heater has unity power factor.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

As far as I can determine, Rheem uses TRIAC cycle-by-cycle phase angle control to modulate the power level delivered to the heater using temperature sensor feedback. They justify this based on the belief that all the other devices in the system should be filtered from the conducted interference they produce by design. Their phase angle control puts the interference spectrum in the high-frequency bands, so that reasoning makes sense. Another consideration is that for fast response the heater doesn't have a huge amount of thermal mass/ capacity, thereby requiring a higher frequency phase angle approach to power modulation so as to make the temperature fluctuations imperceptible to the user.

AI links to this Schaffner EMI filter as a possible solution, 45A and 250V :

There are others, but more people sell and use this manufacturer.

It should be available on eBay for about the same price $150.00. Datasheet is here:

The hard part is the CISPR 17 Filter Attenuation curves. Schaffner has a write-up here:

The water heater is a non-linear 6-ohm, and suprisingly the typical line feed is asymptotic to 50-ohm at high frequency ( so say the LISN practitioners). My guess is the CISPR 17 50/50 asymmetric curve applies, Curve B. And then, based on their write-up of the CISPR 17, the actual application should do no worse than 20dB, making the attenuation at the troublemaking frequencies nearly 60 dB.

Rheem is one of the better manufacturers. They use a sophisticated heat exchanger to maximize the heat transfer and response time. There are many more considerations you're not aware of ranging from engineering, science, available technology, lifetime of components under myriad operating conditions ( not least of which is water chemistry), market acceptance, profits, etc., etc. In the U.S. right now, you have electric resistance, gas flame, heat pump, solar, and that's about it.

If you go heat pump, get the dual with electric resistance heater backup.

NIST just patented a design for a heater that maintains water in the tank at 160F so as to kill Legionella, and then cools it by about 40F as it enters the external pipes by exchanging outgoing water with incoming cold water.

But we haven't been told how Rheem went from mains AC to to the current that they put through their resistive heaters. It won't have been through a 50Hz mains transformer - higher frequency transformers are much cheaper and more compact.

The power factor of the resistive heater isn't going to be all that relevant.

Intelligent plumbers have been warned off?

Absolutely. Screwing up light dimmers is vandalism. Sticking on arbitrary inductance and seeing if it makes a difference is a perfectly respectable way of starting to get to a better engineered solution, and if it does stop the light dimmers from going beserk without needing all the work a a well-engineered solution would take, it is a cheap way to get there.

There's a lot to be said for a bit of quick and dirty tinkering to get a clearer idea of what's going on. It works a lot better if you know what you are doing and have the kind of diagnostic tools that let you see how much difference your tinkering has made, but you can get lucky.

Bill Sloman snipped-for-privacy@ieee.org Wrote in message:r

I have an isobar power tap that I use for such occasions. Have been told in the past, that's not going to work. But it works like s charm every time.

Cheers

Apparently they use a Triac so that they have full control over the power fed into the heaters. The control system seems to attempt to make a kind of PID control loop to keep a constant temperature no matter what the flow rate is.

That does explain the large spikes, potentially one every mains half cycle. That kind of control system really needs a lot more than a ferrite to reduce spurious emissions.

Good points! But for tankless heating, the conventional PID is too sluggish to regulate temperature against potential inlet flow rate and temperature changes that can occur abruptly.

Feed-Forward Calculation (The Bulk Adjustment): Internal flow-rate sensors and inlet temperature thermistors instantly measure real-time changes. Before the water even reaches the heating elements or heat exchanger, the board uses a feed-forward equation (Power = Flow × Δ T × Constant) to calculate roughly 90–95% of the raw energy required to meet the set point.

Once the feed-forward mechanism gets the temperature close to the target, the PID feedback loop steps in. It continuously samples the output temperature sensor and dynamically adjusts or "trims" the remaining percentage of power to account for residual errors or system inefficiencies, holding the target output precisely.

It's all digital.

It can be. That used to be an expensive option. Sampling takes time and imposes a delay, and doing it quickly enough used to be even more expensive.

I think the cheapest ARM microcontroller is about 8 cents. It can probably compute fast enough to handle running water.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

A scope is the tool of choice for that, then you can see what you are dealing with. Would not be surprised if voltage drop is part of the problem though, upsetting other users on the ac line.

All tankless heaters have minimum flow rates that must be detected before they activate the heater. The most common limit is around 1.5 GPM, although some are higher, some are a bit lower at 1 GPM. So flow rate measurement is a fundamental capability they all have anyway.

If you hang an inductive reactance at the input to limit the interference, you will almost certainly have to rework the thyristor snubber network, or face serious consequences ranging anywhere from component failure to loss of turn-off commutation (not good at all).

Since it's a 40-Amp circuit, the filter solution will be pricey. But a $150 self-contained and NEC compliant device wired between the line and heater is cheap compared to what a false-economy alternative solution ends up turning into, and the performance is much better.